cognos transformer step by step guide

IBM Cognos Business Intelligence Series 7IBM Cognos Series 7 PowerPlay Transformer

Version 7.5

Step-by-Step Transformer

Product InformationThis document applies to IBM Cognos Series 7 PowerPlay Transformer Version 7.5 and may also apply to subsequent releases. To check fornewer versions of this document, visit the IBM Cognos Information Centers (http://publib.boulder.ibm.com/infocenter/cogic/v1r0m0/index.jsp).

CopyrightLicensed Materials - Property of IBM© Copyright IBM Corp. 1996, 2010.US Government Users Restricted Rights – Use, duplication or disclosure restricted by GSA ADP Schedule Contract with IBM Corp.IBM, the IBM logo, ibm.com, and Cognos are trademarks or registered trademarks of International Business Machines Corp., in many jurisdictionsworldwide. Other product and service names might be trademarks of IBM or other companies. A current list of IBM trademarks is available onthe Web at www.ibm.com/legal/copytrade.shtml.Microsoft, Windows, Windows NT, and the Windows logo are trademarks of Microsoft Corporation in the United States, other countries, or both.UNIX is a registered trademark of The Open Group in the United States and other countries.

http://publib.boulder.ibm.com/infocenter/cogic/v1r0m0/index.jsp

Table of Contents

Introduction 15PowerPlay Editions 15

Chapter 1: Use the Interface 19Work with the Dimension Map: Overview 19

Add or Delete a Dimension 20Add or Delete a Level 20Show Measure or Data Source Scope 21Show Category Counts 22

Work with Diagrams: Overview 22Connect a Category to Descendants 23

Customize Menus 23Set Up a Launch Menu Command 25

Customize Toolbars and Buttons 25Show or Hide Toolbars 25Create a Custom Toolbar 26Customize Toolbar Buttons 26Set Up a Launch Button 27

Distribute Custom Menus and Toolbars 28

Chapter 2: Get Your Data into the Model 29Create a New Model 30Import an Existing Model in MDL Format 31Add a Data Source to an Existing Model 31Modify SQL 32Re-order the Columns and Data Sources 32Set Up Currency Conversion: Overview 33

Create a Base Currency Table 34Create a Euro Currency Table 36Enable Currency Conversion 39Create or Update a Currency Table Manually 39

Use Multiple Data Sources in Your Model: Overview 40View the Scope of a Data Source 40

Control When Transformer Reads the Source Data 43Use Local Data Sources: Overview 44

Use Spreadsheet Data Sources 44Use PowerHouse Data Sources 47Use Local Databases as Data Sources 48Use Text Data Sources 49Use .iqd Files to Access Server Data 50

Define Columns: Overview 51Define Columns in Fixed-Field Text Data Sources 52Scale Input Values 53Set the Degree of Detail for Dates 54

Licensed Materials – Property of IBM3© Copyright IBM Corp. 1996, 2010.

Use Monthly or Quarterly Time Arrays 55Specify Date Categories in Two Centuries 57

Chapter 3: Map Your Data into Dimensions 59Make a Preliminary Design with AutoDesign 60Create a New Dimension 60Define a Calculated Column 61Add Levels to a Dimension: Overview 62

Add Source Levels to a Dimension 63Add Manual Levels to a Dimension 63Create an Alternate Drill-down Path 65Create Categories Manually 67Create Calculated Categories 68Define Category Sets for Calculated Categories 70Order Categories Within Levels 71Order Categories Using a Global Preference Setting 72Create Unbalanced Hierarchies Within a Dimension 73Define a Scenario Dimension and a Default Opening Level 74

Set Up the Time Dimension: Overview 74Define the Date Input Format 76Use the Date Wizard 76Create a Time Dimension Manually 77Set Up Fiscal Years, Quarters, and Months 79Set Up Calendar and Fiscal Years Within a Single Time Dimension 79Control How Transformer Splits Weeks 81Set Up Lunar Time Periods 82Control When Transformer Adds Extra Weeks for Lunar Time Periods 83Limit the Range of Dates Included in the Model 84Generate Date Categories 85Format Date Values 85Set Up a Custom Time Dimension 85

Set Up Relative Time Categories: Overview 87Set the Current Period Automatically 88Set the Current Period Manually 88Measure Past Performance for a Single Time Period 89Measure Past Performance To Date 90Measure Past Performance Over Several Time Periods 91Set Up Future Relative Time Projections 92Compare Monthly Performance in Different Time Periods 94

Set Up Special Categories: Overview 94Create a New Special Category 95Create Special Category Levels 96Disable Rollup on Special Categories 96

Generate Categories: Overview 97Deal with Orphans Of Categories 97Handle Early and Late Dates 97Handle Invalid Dates 98Use a Specific Data Source to Generate Categories 98Set Up a Manual Level for Unknown Categories 99

4 IBM Cognos Series 7 PowerPlay Transformer

Table of Contents

Check the Model for Potential Problems 100Prevent New Categories from Being Added to a Dimension 100

Chapter 4: Identify the Measures You Want to Analyze 101Add a Regular Measure 101Define a Measure Folder 102Define a Measure that Counts Categories 103Define a Calculated Measure 104Define an If-Then-Else Calculated Measure 105Control When Measures Are Calculated 105Display Missing Values 107Reverse the Sign in Financial Models 107Control How Measures Roll Up: Overview 108

Set a Regular Rollup Function for Measures 108Create Cubes with External Rollups 109Set a Time State Rollup Function for Measures 111Ignore Null and Missing Values in Specified Time State Rollups 113Set Regular and Time State Rollup Together 113

Allocate Measures: Overview 115View Measure Scope 116Set Dimension Allocation 117Set Level Allocation 118Set Category Allocation 119Suppress Allocation 120

Consolidate Data: Overview 120Set a Duplicates Rollup Function for Measures 122Perform Consolidation 124

Set Up Drill-through Targets: Overview 124Add or Remove Drill-through Targets 126Add Drill-through Targets to UNIX Cubes 127

Chapter 5: Define and Create PowerCubes 129Define a PowerCube Manually 129Define PowerCube Groups 130Define a Time-Based Partitioned Cube 130

Slowly Changing Dimensions 134Customizing a Time-Based Partitioned Cube 136

Create a Multi-Level Time-Based Partitioned Cube 137Add or Remove Member Cubes 137

Exclude Measures from a PowerCube 138Omit Dimensions from a PowerCube 138Set a Password for a PowerCube 139Automatic Checking for Open Cubes 140Use Alternate Data Sources for Cube Creation 140Create Compressed PowerCubes 141Create a Test Cube 142Update the PowerCube Metadata 142Check PowerCube Status 143Customize Cube Content with Views: Overview 144

Create a Dimension View 145

Step-by-Step Transformer 5

Table of Contents

Omit Categories Using Suppress 146Omit Categories Using Cloak 147Omit Categories Using Exclude 148Omit Descendant Categories Using Summarize 150Omit Categories Using Apex 150

Set Up a Protected PowerCube: Overview 152Enable User Security for a Model 153Update Authentication Information 154Set Up User Class Views 154Block Total Values for Parent Categories with Excluded Children 155Combine User Class Views with Dimension Views 156

Chapter 6: Perform Model and PowerCube Maintenance 157Recover a Failed Model 158Match Model and Source Columns 159Move Categories When Source Data Changes 160Delete Inactive Categories 161Set Up Placeholder Categories for Cube Groups 161Optimize Performance: Overview 162

Update Your Cubes Incrementally 163Recommendation - Keep Models Populated if Category Codes are Not Unique 165Recommendation - Periodically Clean up Your Model to Optimize Cube Processing 165Use Multiple Data Sources 167Consolidate Data in PowerCubes 167Refresh Category Information 168Devising an Optimal Partitioning Strategy 168Create Multifile PowerCubes 174

Chapter 7: Use Transformer in a Client-Server Environment 177

Chapter 8: Use Architect with Transformer 179Use an Architect Package as a Transformer Data Source 179Create a Transformer Model from an Architect Package 179Add an Architect Package to an Existing Transformer Model 180Metadata Explorer Overview 181Design Your Transformer Model Using the Metadata Explorer 182Filter Your Architect Data Source in Transformer 183Compare the Show Scope Function in Architect and Transformer 184Share Category (General Tab, Category and Special Category Property Sheets) 186

Chapter 9: Reference 187Cognos.ini and <RenditionName>.ini Files 187

Steps 188Trnsfrmr.ini File 190Rendition.cfg File 191The Log File 191

Appendix A: Additional Reference 195Creating Drill-Down Paths 195Defining Scope 195Quotation Marks in Data 196


Table of Contents

Client-Server Model Synchronization 197Server Processes 198Warning and Error Messages 198PowerCube Status Filters 198Data Source Types 199Field Delimiters 201Character Sets 201Isolation Levels in .IQD Files 202UNIX Paths 203Date Input Formats 203Date Format Codes 204Inclusion 205Allocation Types 207Partial and Extra Weeks 207Relative Time Settings 208Degree of Detail 212Level Uniqueness 212Date Functions 213PowerCube Optimization Methods 215Rollup Functions 215Expression Editor 219Search Wildcards 219Category Actions in Diagrams 220

Exclude 221Cloak 222Suppress 222Apex 222Summarize 222

Union of User Class Views 222Scenario Involving Exclude 223Scenario Involving Apex 225Scenario Involving Omit 227Scenarios Involving Parent-Child Relationships 228Scenarios Involving Multidimensional Relationships 230

Alternate Drill-Down Paths in Time Dimensions 231Dimension Views in Alternate Drill-Down Paths 232Current Period 233How Data Sources Relate to the Dimension Map 234Uniqueness Issues When Drilling Cube to Cube 236Level Properties 237

Appendix B: Functions 239change(cat_code, cat_code|set |level) 239percent-growth(cat_code, cat_code|set |level) 239share(cat_code, cat_code|set |cat_code, set|set|level) 240Parameters 240

Appendix C: Command Line Options 241Command Line Options List 242

-a <data_source> 244


Table of Contents

-c 245-cs .mdl_file 245-D<preference_var>=<setting> .py?_file |.mdl_file 246-e .py?_file |.mdl_file 246-F<preference_file> .py?_file |.mdl_file 247-i .py?_file 247-k<signon>=<userid>/<password> .py?_file |.mdl_file 248-m .mdl_file 249-n<display_state> .py?_file |.mdl_file 250-nologo .py?_file |.mdl_file 250-o .py?_file |.mdl_file 250-p .py?_file 250-r<log_level> .py?_file |.mdl_file 251-s .py?_file |.mdl_file 251-t<category_code> .py?_file |.mdl_file 252-u<powercube_name> .py?_file |.mdl_file 252-v data_subset_number 253-x .mdl_file 253- (dash) 253

Appendix D: Proven Practices in Model and System Design 255Recommendation - Analyze Your Requirements and Source Data 255

Identify Your Security Needs 255Assess Your Source Data 256

Recommendation - Build a Prototype 259Identify Measures 259Specify a Time Dimension 259Select the Data to be Modeled (Sales Example) 259Create and Refine Your Model 260Diagnose Design Problems 262Review the Resulting Reports with Users 263

Recommendation - Manually Partition Cubes that do not Support Auto-partitioning 264Recommendation - Test Alternatives for Improving Build Times for Large PowerCubes 268

Example - Explore the Test Model 268Gains from Multiprocessing on a Faster Transformer Server 269Gains from Adjusting the Memory Allocation 270Gains from Reconfiguring the Hard Drive 271Gains from Allocating Sufficient Space for the Temporary Files 271Gains from Optimizing the Windows Environment 272Gains from Optimizing the UNIX Environment 273Gains from Redistributing Files 273Gains from Optimizing Gateway Settings 274Analyze Processing Bottlenecks Using the Transformer Log File 275Keeping Model and Cube Sizes Within Practical Limits 276

Performance Improvement Case Studies 276Recommendation - Uncloak the Primary Drill Category 276Recommendation - Enable Multiprocessing 277Recommendation - Use Time-based Partitioning 278


Table of Contents

Recommendation - Resolve Uniqueness Problems in the Source Before Moving PowerCubesinto IBM Cognos 8 279Removing Missing Values that Cause Uniqueness Problems 280Resolving Uniqueness Problems in the Source 280Retaining the Uniqueness Designation when Categories are Moved 281Preventing New (Non-unique) Categories from Being Added 281

Appendix E: General Troubleshooting 283With the IBM Cognos 8 Web Studios 283

Choosing Only One Language When Installing IBM Cognos 8 OLAP Modeling and MobileAnalysis 283

Providing Stricter Access Controls for PowerCubes Used in IBM Cognos 8 283Not Yet Optimized IBM Cognos PowerCubes May Open Slowly in IBM Cognos 8 283

With Transformer on Windows 285Preventing Errors When Model Calculations Use Double Quotation Marks 285Empty Results When Querying a Microsoft Access Database 285Access Time Increases with Incremental Updates 286Add New Partition Levels to Existing PowerCubes 286Retain Sort Order in Time-Based Partitioned Cubes 286Allocations Aren't Working Correctly 286Alternate Drill-downs and Level Uniqueness 287Alternate Drill-downs and Partitioning 287Can't Join the Source Data Files in My Data Sources List 288Category Generation Is Too Slow 288Dimension Contains Orphans Of Categories 288Dimension Contains Unbalanced Hierarchy of Categories 288Incremental Update Failed for One or More PowerCubes 288Invalid Dates Generated in the Time Dimension 289Level Uniqueness Problems 289Measures Have No Values in Some Dimensions 291Relative Time Calculations Aren't Evaluated Correctly 292Source Data Doesn't Match the Data Source in the Model 292Allocation Tab Is Disabled for All Objects in a Model 292Inconsistent Numeric Output After Migrating Models Created Using Native dBase Data

Sources 293Transformer Seems to Ignore a Measure's Decimal Values 293Oracle Database Schema Definition Modifies Numeric Category Labels 294Unknown Signon Appears in Signons List with .iqd File 294User Class Security Isn't Working 295When or Why Should I Use Multiple Data Sources? 296ObjectStore VirtualAlloc Failure Prevents Startup of Transformer 297Allocation Appears Incorrect in Time-based Partitioned Cubes if a Cube is Updated with

Data from Another Period 297Unable to Import IBM Cognos 8 Data that Contains TEXTBLOB, CLOB, or BLOB Data

Types 298Insufficient Cache Space for Proper Cube Processing 298Unable to Connect to the Database 298

With Transformer on UNIX 299Client-Server Connection Is Not Working 299


Table of Contents

Can't Create a Model Using Transformer on UNIX 299With PowerPlay Client Applications 300

Production Reports No Longer Work 300Dates and Numeric Values Not Formatted as Expected 300Inadequate Query Performance with Partitioning in Place 300PowerCube Access from PowerPlay Clients Is Too Slow 301Summaries Don't Match Individual Cell Contents 301

Appendix F: Error Messages 303Message TR0104 303Message TR0105 303Message TR0106 303Message TR0107 304Message TR0108 304Message TR0109 305Message TR0110 305Message TR0111 305Message TR0112 305Message TR0113 306Message TR0114 306Message TR0116 306Message TR0117 306Message TR0118 307Message TR0128 308Message TR0131 308Message TR0132 308Message TR0133 309Message TR0137 309Message TR0149 310Message TR0151 310Message TR0202 310Message TR0203 310Message TR0205 311Message TR0206 311Message TR0207 311Message TR0208 311Message TR0209 311Message TR0210 312Message TR0214 312Message TR0215 312Message TR0217 313Message TR0301 313Message TR0303 313Message TR0404 313Message TR0407 313Message TR0408 314Message TR0412 314Message TR0420 314Message TR0423 314


Table of Contents

Message TR0476 315Message TR0500 315Message TR0501 315Message TR0502 315Message TR0503 316Message TR0504 316Message TR0505 316Message TR0507 317Message TR0508 317Message TR0510 317Message TR0514 317Message TR0515 318Message TR0518 318Message TR0519 318Message TR0523 318Message TR0524 319Message TR0525 319Message TR0528 320Message TR0534 320Message TR0535 320Message TR0536 320Message TR0538 321Message TR0540 321Message TR0541 321Message TR0552 322Message TR0605 322Message TR0606 322Message TR0607 323Message TR0613 323Message TR0621 323Message TR0623 324Message TR0624 324Message TR0700 324Message TR0701 325Message TR0702 325Message TR0705 325Message TR0713 325Message TR0716 326Message TR0743 326Message TR0745 326Message TR0749 327Message TR0750 327Message TR0751 327Message TR0752 327Message TR0753 327Message TR0800 327Message TR0802 328Message TR0803 328Message TR0804 329


Table of Contents

Message TR0805 329Message TR0806 329Message TR0807 329Message TR0808 329Message TR0809 330Message TR0810 330Message TR0811 330Message TR0812 331Message TR0813 331Message TR0815 331Message TR0816 331Message TR0817 332Message TR0900 332Message TR0904 332Message TR0906 332Message TR0907 333Message TR0914 333Message TR0917 333Message TR0919 333Message TR1003 334Message TR1005 334Message TR1100 334Message TR1101 334Message TR1102 334Message TR1106 335Message TR1109 335Message TR1304 335Message TR1307 335Message TR1308 336Message TR1309 336Message TR1310 336Message TR1312 336Message TR1319 337Message TR1320 337Message TR1340 337Message TR1502 338Message TR1503 338Message TR1601 338Message TR1700 338Message TR1703 339Message TR1900 339Message TR1903 339Message TR1907 340Message TR2000 340Message TR2001 340Message TR2002 341Message TR2306 341Message TR2307 341Message TR2308 341


Table of Contents

Message TR2312 342Message TR2313 342Message TR2314 342Message TR2316 343Message TR2317 343Message TR2318 343Message TR2319 344Message TR2320 344Message TR2321 345Message TR2322 345Message TR2323 345Message TR2324 345Message TR2325 346Message TR2326 346Message TR2502 346Message TR2503 347Message TR2504 347Message TR2505 347Message TR2506 347Message TR2507 348Message TR2508 348MessageTR2509 348Message TR2510 348Message TR2600 349Message TR2601 349Message TR2605 349Message TR2606 349Message TR2608 350Message TR2700 350Message TR2701 350Message TR2702 351Message TR2703 351Message TR2704 352Message TR2705 352Message TR2706 352Message TR2707 352Message TR2708 353Message TR2710 353Message TR2711 353Message TR2712 354Message TR2713 354Message TR2714 354Message TR2715 354Message TR2716 355Message TR2717 355Message TR2718 355Message TR2719 356Message TR2720 356Message TR2723 356


Table of Contents

Message TR2724 356Message TR2725 357Message TR2726 357Message TR2727 357Message TR2728 358Message TR2729 358Message TR2731 358Message TR2732 358Message TR2733 359Message TR2734 359Message TR2735 359Message TR2736 359Message TR2737 360Message TR2740 360Message TR2800 360Message TR2801 360Message TR2802 361Message TR2803 362Message TR2804 362Message TR3006 362Message TR3010 363Message TR3014 363Message TR3101 363Message TR3311 363Message TR3439 364Message TR3505 364Message TR3619 364

Glossary 365

Index 395


Table of Contents

Introduction

This document presents procedures, examples, best practices, troubleshooting tips, and other

background information to help you create PowerCubes in Transformer, the OLAP modeling

component used with PowerPlay.

If you are new to Transformer, we recommend that you take the multimedia Quick Tour to become

familiar with how PowerPlay works. Follow the tutorials in the Discovering PowerPlay and Dis-

covering Transformer books, to gain a basic understanding of PowerPlay and its modeling compon-

ent. You can then use this document to help you design and fine-tune your Transformer models.

PowerPlay EditionsDepending on your PowerPlay edition, portions of some topics may not apply to you. Click here

(p. 15) for details on the features of Transformer in the various PowerPlay editions.

Features and Editions

The features and functionality of Transformer depend on your PowerPlay edition. You can't open

a model if your version of Transformer does not support one or more of the features included in

that model.

For example, in Personal Transformer, you can't open a model that includes cube groups or access

restrictions based on user class.

To determine your edition, click About IBM Cognos PowerPlay Transformer (Help menu).

Personal Transformer (Series 7 PowerPlay User and IBMCognos 8 Business IntelligenceMobile Analysis 8.1 Windows)

You can

● store a cube in Local format

● use multiple data sources for your model

● use a variety of source types

● create a single cube based on a model

● use MDL to maintain models

The interface has the following characteristics:

● The Access Manager Administration tool is not available; you must use Configuration Manager

to configure your default authentication source.

● There is no Server menu.

● The Model property sheet has no Server or Authentication tab.


● The Source tab (Data Source property sheet) has no Data Source Location box and the General

tab has no Enable Multi-Processing check box.

● The PowerCube property sheet has no Dimensions or Cube Group tabs.

● The Processing tab (PowerCube property sheet) has no Cube Creation or Processed boxes and

the Output tab has no Password box.

● The Diagram menu does not have the Cloak, Apex, or Summarize options and the User Class

tab in the diagram is not available.

Transformer for Windows (Series 7) and IBM Cognos 8 Business Intelligence OLAPModeler 8.1 (Windows)

You can perform all the functions possible with Personal Transformer.

In addition, you can

● create multiple cubes (groups) based on a model

● create cubes tailored for widespread distribution by using dimension views and user class views

● create cube groups

● take advantage of your computer's multiple processors to improve retrieval performance

The interface has the following characteristics:

● There is no Server menu.

● The Model property sheet has no Server tab.

● The Processing tab (PowerCube property sheet) has no Processed box.

● The Source tab (Data Source property sheet) has no Data Source Location box.

Transformer for UNIX (Series 7) and IBMCognos 8 Business Intelligence OLAP Builder8.1 (UNIX)

These editions support the processing, storage, and maintenance of cubes from the UNIX command

line using the rsserver executable.

In addition to including all the documented interface items, the Series 7 Transformer for UNIX

server supports automatic synchronization between Windows and UNIX server operations. For

more information, see the Transformer for UNIX guide.

Note: Some Asian languages are not supported by IBM Cognos Impromptu, IBM Cognos Impromptu

Web Reports, and IBM Cognos Visualizer.

Finding information

To find the most current product documentation, including all translated documentation, access

one of the IBM Cognos Information Centers at http://publib.boulder.ibm.com/infocenter/cogic/

v1r0m0/index.jsp.

You can also read PDF versions of the product release notes and installation guides directly from

IBM Cognos product disks.


Introduction



Samples disclaimer

The Great Outdoors Company, GO Sales, any variation of the Great Outdoors name, and Planning

Sample depict fictitious business operations with sample data used to develop sample applications

for IBM and IBM customers. These fictitious records include sample data for sales transactions,

product distribution, finance, and human resources. Any resemblance to actual names, addresses,

contact numbers, or transaction values, is coincidental. Other sample files may contain fictional

data manually or machine generated, factual data compiled from academic or public sources, or

data used with permission of the copyright holder, for use as sample data to develop sample

applications. Product names referenced may be the trademarks of their respective owners. Unau-

thorized duplication is prohibited.


Introduction


Introduction

Chapter 1: Use the Interface

Property Sheets

Each object defined in a Transformer model (such as a level in the dimension map, a category in

the category viewer, or an item in a list) has properties that you control or specify from that object's

property sheet.

To open the property sheet for a model object, you can

● select the object, and then from the Edit menu, click Properties

● select the object, then right-click and select Properties

● double-click on the object

You can get What's this? help for any element in a property sheet by clicking the Question Mark

in the top right-hand corner of the dialog box (not the toolbar) and then clicking the item for which

you want help .

Lists

Most objects that you work with in Transformer appear in lists. You can use these lists to add,

move, or delete model objects, or to quickly open the associated property sheet. The Data Sources,

Measures, and PowerCubes lists appear below the dimension map when you create a new model.

If the following are enabled for your model, they may also be selected from the View menu:

● User Classes

● Signons

Pop-up Menus

Nearly every menu command is available from a pop-up menu, which you can open by selecting

an object and right-clicking. You can open pop-up menus for all objects in lists, diagrams, and the

dimension map. You can also open pop-up menus from within certain property sheets, the

Dimensions tab (PowerCube property sheet), for example. If you choose multiple objects of the

same type (for example, multiple categories in the category viewer), you can apply pop-up menu

commands to several objects at once.

Work with the Dimension Map: OverviewThe dimension map is a table that shows all dimensions in the model, and the levels they contain.

It is the working area where you see your model taking shape. Use it to plan the structure of your

multidimensional cube, including

● a time dimension that contains the relevant divisions in the calendar or fiscal year, such as

Quarter, Week, and Day, generally derived from a single column in your data source


● alternate drill-down paths, to support analysis by different attributes, such as product type or

supplier

● manually added levels (for example between Region and State) so you can more accurately

allocate measures

● associations between levels and transaction values (such as Quantity Sold) for multiple data

source models

The dimension map consists of a dimension line, where you add dimensions, and levels for each

dimension.

Tip: Use the AutoDesign command (Tools menu) to build a preliminary dimension map based on

your source data.

Add or Delete a DimensionYou can use the dimension map to add or delete dimensions in your model. When you use the

dimension map, the selected part of the dimension line becomes the insertion or deletion point. To

change the position of a dimension, drag it to another location on the line.

You can drag a column from the Data Sources list onto the dimension line or Dimensions list to

create a new dimension. The new dimension contains a single source level based on the column.

Steps to Add a New Dimension

1. On the dimension line, select the dimension to the right of where you want the new dimension

to go, or the empty space to the right of the last dimension.

2. From the Edit menu, click Insert Dimension.

Steps to Delete a Dimension

1. On the dimension line, select the dimension that you want to delete.

2. From the Edit menu, click Delete Dimension.

Note: If the dimension contains categories, you are prompted to confirm the deletion.

Add or Delete a LevelTo add a level to a dimension in your model, you can use the Insert command (Edit menu), or drag

a column from the Data Sources list to the dimension map.

If you use the menu command, a manual level is automatically created. If you want to change the

manual level to a source level, open the Level property sheet and set the Associations.

If you insert a level by dragging the corresponding column to the dimension map, a small box

indicates where it will be dropped. In this example, the box indicates that the new level will appear

below the level Brand.



This box indicates that the new level will appear between the existing levels Line and Item.

This box indicates that an alternate drill-down path will be created with Item as the convergence

level.

Steps to Add a Level

1. In the dimension map, select the level above which you want the new level to appear.

2. From the Edit menu, click Insert Level.

Steps to Delete a Level

1. In the dimension map or category viewer, select the level that you want to delete.

2. From the Edit menu, click Delete Level.

Show Measure or Data Source ScopeIf you use multiple data sources or measure allocation in your model, you can check the relationships

between various levels in the dimension map by using the Show Scope option. This applies to both

data sources and their measures. For more information, see "View the Scope of a Data

Source" (p. 40) and "View Measure Scope" (p. 116). You can also open the Scope dimension map

from the pop-up menu for an individual data source or measure.

Steps

1. With the dimension map open, click Show Scope (Edit menu).

2. Click the Measures or Data Sources option and select the measure or data source whose scope

you want to see.

Tip: You can perform the same actions from the Scope dimension map as you can from the

default dimension map, such as adding a new dimension.

3. To change from the Scope dimension map back to the default map, click Show Scope a second

time.



Show Category CountsWhen you select Show Counts (Edit menu), the dimension map changes to show

● the number of regular categories in each level, which appear within parentheses, following each

level name

● the number of other categories in each dimension (that is, drill categories, root categories,

special categories and categories in subdimensions), which appear within parentheses, below

the dimension names in the dimension line

● the total number of categories in each dimension, which appear within parentheses, beside each

dimension name on the dimension line

You can also show category counts by using the Show Category Counts button on the toolbar.

Steps

1. With the dimension map open, from the Edit menu, click Show Counts.

The dimension map shows the number of categories in each level, and the totals for each

dimension.

2. To hide category counts, click Show Counts again.

Work with Diagrams: OverviewWhen you select Show Diagram in the Diagram menu, or click its toolbar

● the left pane shows all the dimensions and dimension views in your model, in a tree structure

● the right pane, or category viewer, shows all the levels and categories in the selected dimension

● if user classes are enabled for your model, the User Class tab shows a list of authenticated user

classes enabled for the model, in a Series 7 namespace-type tree structure

You can access the main menu, toolbar buttons, and pop-up menus from any pane and use these

controls to fine-tune your model. For example, you can

● expand or collapse folders, to show or hide details

● zoom in and out, with the font size changing automatically to suit the degree of magnification

● access the associated property sheet by double-clicking an item

● add or delete one or more categories or levels

● use the Show Reference command to show the origins of a level

● move one or more categories, provided the Order By feature is not specified for the level

● exclude or suppress a level

● define a set of levels or categories, and apply one or more operations or calculations to them



Notes

● If you add a special or manual category with descendants, ensure that they do not conflict with

other categories in the model, or destroy level uniqueness. See "Connect a Category to Descend-

ants" (p. 23).

● For additional task-related information, see "Add Source Levels to a Dimension" (p. 63).

● To learn the basic modeling skills, complete the introductory tutorials in the Discovering

Transformer tutorial.

● Depending on your PowerPlay edition, portions of this topic may not apply to you. For more

information about the features of Transformer in the various PowerPlay editions, see the Features

and Editions section in the Welcome chapter.

Connect a Category to DescendantsWhen Transformer generates categories for your model, it automatically connects them to their

descendants. If you add a special category, you can manually connect one or more descendant cat-

egories to it.

In the category viewer, when you drag the mouse pointer to link a descendant to its parent category,

the connecting line appears dynamically, showing you where a valid connection can be made.

You can connect categories to descendants at levels above or below the existing category level, but

not to descendants at the same level.

Steps

1. In the category viewer, position the mouse pointer over the right side of the root category, or

any manual or special category that you want to link descendants to.

2. When the pointer changes to a cross-hair, click the left mouse button, drag a connection to the

right, and then release the mouse button when it is no longer positioned over any existing cat-

egories (for a brand new descendant category), or when it is positioned on the descendant.

3. Set the properties for the newly connected descendant as required.

Customize MenusYou can create your own menus and add the commands you need. When you change default menus

or create new ones, the changes are stored in an .mnu file, which is updated each time you make a

change.

Customized menus appear only when a report is open.

Steps

1. From the Tools menu, click Customize.

2. Click the Menus tab.

A list of menu commands appears in the Available Commands box. The Menu Layout box

shows the current menu structure.



3. You can do any of the following.

ActionGoal

Drag what you want to add from the Available

Commands box to the Menu Layout box.

Add menus, submenus, menu separat-

ors, or menu commands

In the Category box, click the name of the menu.

For example, click File to show only those com-

mands found on the File menu.

Show only those menu commands

found on a specific menu

In the Menu Layout box, drag menu commands

from one menu to another.

Rearrange menu commands

In the Menu Layout box, click the menu or com-

mand and then click Remove.

Delete a menu or menu command

In the Menu Layout box, click the menu com-

mand. Under Item Properties, in the Name box,

type a new name.

Change the name of a menu command


mand. Under Item Properties, in the Shortcut box,

enter a new combination of keys. If a combination

of keys is currently assigned to another menu

command, a message will appear in the message

pane that the shortcut is in use.

Change the shortcut key combination

of a menu command


mand. Under Item Properties, in the Name box,

type an ampersand (&) and the name of the menu

command. For example, &New makes N the

accelerator key.

Create a menu accelerator key

4. Click OK.

Tips

● You can add a separator line between groups of related commands to organize a menu.

● To restore the default menus, click the Reset button.



Set Up a Launch Menu CommandYou can add a launch command to any menu. You can configure each launch menu command to

run a different program or macro.

Steps


2. Click the Menus tab.

3. Click New.

The New Launch Item dialog box appears.

4. In the Name box, type a name for the launch menu command.

5. In the Description box, type a tooltip for the menu command.

6. In the Type box, click Macro or Application.

7. In the Command box, browse to the location of the macro or application.

8. In the Parameters box, enter any parameters you want to add to the macro or program.

9. If you want to minimize the current window when you run the program or macro, select the

Minimize Application Before Launching check box.

10. Click OK.

Tip: To edit a launch menu command, click it in the Menu Layout box, then click Modify.

Customize Toolbars and ButtonsYou can show, hide, create, delete, and modify toolbars. Any changes you make are stored in an

initialization file on your computer, for the current user.

Show or Hide ToolbarsYou can show or hide toolbars so that you see only what you need.

Steps


2. Click the Toolbars tab.

A list of toolbars appears in the Available Toolbars box.

3. Select the check box beside each toolbar that you want to appear.

Tip: You can remove a toolbar from the Available Toolbars box by clicking the toolbar name,

and then clicking Remove.

4. Click OK.

The settings are saved for your computer.



Tip:To show tooltips for each toolbar, select Show Tooltips.

Create a Custom ToolbarYou can create a custom toolbar that contains only the tools you need.

Steps


2. Click the Toolbars tab.

3. Click New.

4. In the Toolbar Name box, type a name for the toolbar.

5. In the Initial Location box, click a default location for the toolbar.

6. Click OK.

The name of the new toolbar appears in the Available Toolbars box. You can now add buttons

to the new toolbar.

Tip:To reset the toolbars to the default settings and delete any custom toolbars, on the Menus tab,

click Reset. You can also reset the toolbars by deleting the tbinfo.ini file or moving it to another

folder.

Customize Toolbar ButtonsYou can change the buttons on a default toolbar or add buttons to a custom toolbar. You can also

remove buttons from any toolbar.

To learn what a button does, select its name in the Available Buttons box, then look under Button

Description.

For example, you are in charge of imports for a governmental department. You regularly distribute

reports to offices in Great Britain, which requires that you convert your reports so that they use

British pounds. You want to customize your toolbar by creating and adding a Currency button.

Steps


2. Click the Toolbar Buttons tab.

3. Drag the buttons you want to add from the Available Buttons box to a toolbar.

Tips:

● To change the selection of buttons use the Category box. For example, click File to show

only those buttons found on the File toolbar.

● To add a space between buttons, add a separator button (found under the Miscellaneouscategory).

4. Drag the buttons you want to remove off an existing toolbar.



5. Click OK.

Tip:To return to the original settings, click Reset on the Toolbars tab.

Set Up a Launch ButtonYou can add up to 64 launch buttons to a custom toolbar. You can configure each button to run

a different program or macro.

For example, you are the manager of a government department. You keep the data source files for

your reports in Microsoft Excel spreadsheets. While reviewing your reports, you want to look at

your data source files. You customize your toolbar to run Excel directly from PowerPlay, Impromptu,

or Transformer.

Steps


2. Click the Toolbar Buttons tab.

3. In the Category box, click Miscellaneous.

The buttons in the Miscellaneous category appear in the Available Buttons box.

4. Click Toolbar Launch Button, and drag it to a toolbar.

5. Double-click the new launch button.

The New Launch Item dialog box appears.

6. In the Name box, type a name for the launch button.

7. In the Description box, type a brief description of what the button does.

The tooltip appears when you pause the pointer over the button.

8. In the Type box, click Macro or Application.

9. In the Command box, browse to the location of the macro or program.

10. In the Parameters box, enter any parameters you want to add to the macro or program.

11. If you want to minimize the current window when you run the program or macro, select the

Minimize Application Before Launching check box.

12. In the Image box, click the image you want to appear on the new launch button.

13. In the Path box, browse to the location of the button image.

14. Click OK.

Tip:To change the settings of a launch button, ensure the Customize dialog box is closed, and

Ctrl+click the launch button and then make changes in the Modify Launch Item dialog box.



Distribute Custom Menus and ToolbarsYou can distribute custom menus and toolbars to other users.

Toolbars are stored in the tbinfo.ini file, and menus are stored in the menuinfo.mnu file in the fol-

lowing location:

<installation_dir>\Documents and Settings\<user_id>\Application Data\Cognos\cern\<filename.exe>.

For example, if you are running Impromptu you would find tbinfo.ini and menuinfo.mnu in the

following location:

<installation_dir>\Document Settings\<user_id>\ApplicationData\Cognos\cern\impadmin

Example

You are the director of a government department that oversees transportation policy. You worked

closely with your systems analyst to build customized menus and toolbars that suit your departmental

needs. You want to distribute these menus and toolbars to others in your department.

Steps

1. Place a copy of the files that contain the settings for the custom toolbars and menus in a location

that users can access.


3. Click Load.

4. Browse to the location of the tbinfo.ini file or the menuinfo.mnu file to be imported.

5. Click OK.

The custom menus or toolbars appear.

Note: Users cannot import a configuration file created in PowerPlay 7.0 RTM and previous versions.

You must run PowerPlay with the old settings and click Save As to save the settings in a new initial-

ization file.



Chapter 2: Get Your Data into the Model

Transformer models contain definitions of dimensions, levels, and measures. They may also contain

dimension views, user classes, user class views, and other authentication-related information.

When "populated" with data drawn from one or more named data sources, models also contain

multidimensional cubes or cube groups that Transformer creates for reporting in Series 7 PowerPlay

or the IBM Cognos 8 Web studios. You can store your models as files with the extension .mdl

(exportable ASCII) or .py? (binary form), where the final character represents the version of

Transformer used to create the model.

Data sources may be structural or transactional:

● Structural data sources contain the columns that define the model structure: the categories in

each dimension. They likely contain many columns, but few rows.

● Transactional data sources contain the columns for the measures to be tracked. They likely

contain few columns, typically only one for each dimension and one for each measure, but

many rows.

Although the resulting model may be inefficient, a source can contain both structural and transac-

tional information.

Guiding Principles

When setting up data sources

● if possible, design your data so that one source provides the structural information for each

dimension

● if your data sources mix transactional and structural data, place the primarily structural ones

first, and optimize processing efficiency by using Generate Categories from the Selected Data

Source option (Run menu)

● ensure that each source contains enough information to generate the categories for a dimension

without database joins, or join queries from separate database tables by using access tools such

as IBM Cognos Series 7 Impromptu or IBM Cognos 8 Framework Manager, before you import

the data into Transformer

Notes

● Models based on multiple data sources are supported by all editions of PowerPlay and Trans-

former.

● In Series 7 Transformer, you may implement password-protected cubes, but other means of

controlling access to information may or may not be supported, depending on your PowerPlay

edition. For more information about the features of Transformer in the various PowerPlay

editions, see the Features and Editions section in the Welcome chapter.


Create a New ModelEvery cube that you access in PowerPlay is based on a model. Creating a new model involves

● specifying the data sources for the model

● defining the dimensions, levels, and measures in a Series 7 model

● defining cube objects that use the content of the model to create cubes (optional)

To create a new copy of an existing model, use the Save As command (File menu). Transformer

prompts you to save a copy of your model in binary (.py?) or ASCII text (.mdl) format. Use the

.mdl format when exporting models, or when you have made a lot of changes since your last save

action, to avoid possible fragmentation problems.

It is good design practice to periodically save your models in .mdl format. Also, you should regularly

use the Check Model command (Tools menu) to help you diagnose and resolve any problems in

your model design.

Steps

1. From the File menu, click New to open the New Model wizard.

2. Enter a name for your model in the Model Name box, click Next, enter the name of your initial

data source in the Data Source Name box, select a Source Type from the list, and click Next.

3. If you plan to control data access by means of user class authentication, select the Include

Security in this Model check box.

4. Depending on your edition and selected data source, Transformer shows different options.

Enter the parameters that correspond to your requirements.

For example, if your source is an Excel database, you must provide the named range of cells

from the Excel worksheet.

5. On the last page of the wizard, select Run AutoDesign to have Transformer create a preliminary

structure for you; or clear this check box to design the model yourself.

6. When you have entered all of the required information, click Finish.

Transformer opens a new model based on your specified source, and shows you information

about the source data:

If you selected the Run AutoDesign option, you see your source columns in the Data Sourceslist, some "best guess" dimensions and levels in the dimension map, and numeric columns in

the Measures list.

If you cleared the autodesign option, you see the Data Sources list, but all remaining windows

are blank.

Note: You can use the Data Source Viewer to display sample data and the SQL from your data

source. Click on the Data Source Viewer button in the toolbar to start the Data Source Viewer.



Import an Existing Model in MDL FormatIf you have been using a previous version of Transformer, you must save existing models in Model

Definition Language (.mdl) format and then import that file directly into Transformer. When the

.mdl file is imported into the current version of Transformer, equivalent definitions are created for

all model objects.

When importing .mdl files from earlier Transformer versions, bear in mind that some features may

not convert. Be particularly careful of user class views based on namespaces administered by Access

Manager, and legacy data that contains special characters, spaces, or quotation marks.

Steps

1. Open the model in the earlier version of Transformer and, from the File menu, click Save As.

2. In the Save as Type box, click Exported Model Files (*.mdl).

Note: You can set the default location in which Transformer saves models by changing the

Models directory in the Directories tab (Preferences dialog box).

3. In the current version of Transformer, open the resulting .mdl file.

4. For production use, save the converted model in binary (py?) format. Otherwise, resave it as

.mdl.

Add a Data Source to an Existing ModelWhen starting a model using the New Model wizard, you can only specify one data source. The

first data source you specify should be a large structural data source containing columns that define

the key dimensions (and hence the structure) of your model. You can then use the Insert Data Source

command to open the New Data Source wizard, and add any other sources required for your model.

Ensure that each data source has a unique name.

Notes

● If you normally work in a server environment, you should create a prototype model locally, to

start. To speed up the design and test phase, use only a portion of your source file. Depending

on your edition, for more information, see the "Using Transformer in a Client-Server Environ-

ment" or "Building PowerCubes on UNIX" sections of the online help.

● If your model is based on more than one data source, you may want to allocate measures to

other levels, based on a constant or another measure in your model. Allocation between data

sources is supported by all editions of PowerPlay and Transformer.

● If you plan to control data access by means of user authentication, ensure that the Include

Security in this Model check box is selected, and that your authentication source is appropriately

configured in Access Manager.

Steps

1. Right-click in the Data Sources list of your model and select Insert Data Source.



2. In the New Data Source wizard, enter the name of the data source in the Data Source Name

box, select a Source Type from the list, click Next and, in the Local Data File box, enter the

name of the source file.

3. Depending on your edition and selected data source, Transformer shows different options.

Enter the parameters that correspond to your requirements and then click Finish.

For example, if your source is an Excel database, you must provide the named range of cells

from the Excel worksheet.

4. Confirm that your selected items appear as expected in the Data Sources list.

Modify SQLSQL, or Structured Query Language, is the standard query language for accessing relational database

information. Modifying SQL can optimize, customize and refine queries.

Modify the SQL in Transformer by using the Data Source Viewer.

Notes

● This feature is only valid for .iqd files.

● After modifying SQL, the relationship between columns in your .iqd file and your model may

be lost if:

● there is an expression in your SQL

● there is an input scale in your .iqd file

● the column names of the items below the END SQL line do not match with those column

names in the SQL

Steps

1. From the View menu, click Data Source Viewer.

2. To open the SQL Modify window, click Modify.

By default, the SQL Modify window opens on the SQL tab.

3. To make changes to the SQL, click in the SQL tab and make your changes.

4. To view your changes, click the Preview tab.

5. To close the SQL Modify window without saving your changes, click Cancel.

6. To save your changes, click OK.

Re-order the Columns and Data SourcesBy default, columns appear in Transformer in the same order as they occur in the data source. You

can re-order columns in the Data Sources list so that related columns appear together, without



affecting the processing of that source. The position of each column in the data row is specified in

the column definition, regardless of where the column appears in the Data Sources list.

If your model uses multiple date columns, you can specify which column sets the current period

and then lock in that setting by clearing the Automatically Set Current Time Period option in the

time dimension property sheet. Alternatively, you can automatically set the current period to the

latest date overall by clearing the Sets the Current Period property for all data sources.

Steps

1. To re-order a column, drag it to the new position in the parent data source.

2. To re-order a data source, drag it to another position in the Data Sources list.

Set Up Currency Conversion: OverviewIf your users need to see reports in more than one currency, you can add one or more data sources

to your model to supply the necessary conversion rates. Transformer maintains the association with

this external data in the Currency Table, and uses the information to make the correct conversions

when cubes are viewed in PowerPlay.

We recommend that as a best practice you always define a default currency for your data within

your Transformer model.

The currency information is stored at the same level of detail as appears in the time dimension of

your model (by default, monthly). However, you can adjust this to a higher level (such as quarterly),

if this meets your users' needs. When your PowerPlay users select the appropriate currency conversion

command, the conversion rate for that time period is applied to the selected measures in the cube.

Converted values are subsequently rolled up as specified in the Measure property sheet. For more

information, see "Control How Measures Roll Up: Overview" (p. 108).

The procedures you follow to set up currency conversion using external data sources depend on

your situation. If you are converting to or from the currencies of European Monetary Union (EMU)

countries, there are extra steps to follow. Also, if you use more than one data source to supply the

conversion rates for the Currency Table, you must ensure that the column names match, so that

the information combines properly in Transformer.

To convert between the currencies of countries that are not members of the European Monetary

Union (EMU), you follow a two-stage process:

● "Create a Base Currency Table" (p. 34)

● "Enable Currency Conversion" (p. 39)

To convert legacy data from one EMU country currency to another, we recommend that you follow

the three-stage Euro triangulation process:

● "Create a Base Currency Table" (p. 34)

● "Create a Euro Currency Table" (p. 36)

● "Enable Currency Conversion" (p. 39)



For simple models, you may want to manually create or update the Base and Euro Tables, instead

of using external data sources to supply the currency conversion rates. See "Create or Update a

Currency Table Manually" (p. 39).

Note:All editions of PowerPlay and Transformer support models created from multiple data sources,

so you can load conversion rates from external files, or manually create them in Transformer

(whichever is easier).

Create a Base Currency TableUnless you are converting to or from the currencies of European Monetary Union (EMU) countries,

the Currency Table consists of one pane: the Base Table. Initially it shows the default currency of

the country specified in the Locale setting of your operating system, such as US dollars. You can

change this <Base default> to any other currency, including the euro. When you load a conversion

table into a populated model, the data appears here.

Base Currency File

This comma-delimited sample currency file contains conversion rates for the first quarter of 1999.

It includes the three mandatory columns (Date, Country Code, and Rate), as well as the optional

Currency Label column.

You can change rates marked with a pencil .

Initially, rates have a default value of 1.0000.

DATE,CURRENCY LABEL,COUNTRY CODE,RATE

19990101,Canadian Dollars,CAN,1.5223



19990101,British Pound,GBR,0.6072



19990101,Japanese Yen,JPN,113.4700



Rates are expressed as the amount by which you multiply a unit of the given currency to buy a unit

of the base currency. In March 1999, 1 US dollar (base currency) cost 1.51 Canadian dollars, 0.60

British pounds, and 112.90 Japanese yen (sample rates, for illustration purposes only).

Notes

● You should clear the Timing options in the property sheets for all currency data sources so that

updates to the Currency Table do not interfere with category generation in the rest of your

model. If you decide to leave all Timing options at their default settings, ensure that the names

of the date columns for the base and euro currencies differ from the name of the column used

to create your time dimension, and from each other. Otherwise, Transformer can't differentiate

between them.



● Check Model issues a warning if your model contains a Currency Table but none of the measures

have currency conversion enabled. You can ignore this message, but currency conversion will

not be available in PowerPlay.

● The Currency Symbols and Decimal Formats that appear in PowerPlay are determined by the

Country Codes. You can override these defaults in the Currency Record property sheet.

Steps to Prepare Base Table Source Files

1. Create one or more source files to populate the Currency Conversion Table in your model.

These files should contain columns for

● the dates to which the conversion rates apply (except for fixed EMU-entry rates)

● the codes for the countries to which the conversion rates apply

● the conversion rates, expressed as the amount by which you multiply a unit of the given

currency to buy a unit of the base currency

● optionally, a label for the currency (if you don't specify a label, it is derived from the

country code)

2. If you have a time dimension that doesn't contain unique levels, you must change it to contain

unique levels.

See "Create a Time Dimension Manually" (p. 77).

3. If categories have not been generated for the time dimension of your model, click Generate

Categories from the Run menu. (Otherwise, you won't be able to see the details in the Currency

Table until you create your cube.)

4. Add a data source to your model for each source file.

See "Add a Data Source to an Existing Model" (p. 31).

Note: If you create the time dimension from more than one data source, the date column names

must match.

5. Double-click each source in the Data Sources list to open its property sheet, and click the

General tab.

Clear the Timing check boxes and the Sets the Current Period check box, or ensure that the conver-

sion date column has a different name from that of the column used to create the time dimension.

Click OK.

Steps to Create a Base Table

1. From the File menu, click Currency Table and select the Use an External Currency Data Sourcecheck box.

2. If you want to change the base currency, right-click the <Base Default> currency. Click Properties

to open the Currency Record property sheet, change the Country Code, and click OK.

Note: The <Base Default> currency is initially defined by the Locale setting of your operating

system.



3. Confirm that the Source Associations window shows the correct Date, Country Code, and

Conversion Rate columns to populate the Base Table. If not, select Modify in the pop-up menu

and make corrections.

4. Repeat Step 3 for the Currency Label column, if present. (This name may be derived from the

Country Code.)

5. Click Load Table.

If your model contains date categories, conversion rates are added as specified and the currency

records are created.

Note: Currency records are also generated when you generate categories or create a cube.

6. Follow the steps in "Enable Currency Conversion" (p. 39).

Create a Euro Currency TableWhen you select the Use Euro Triangulation option to convert to or from the currencies of European

Monetary Union (EMU) countries, the Currency Table appears as two panes. The left pane, or Base

Table, initially shows the default currency of the country specified in the Locale setting of your

operating system, such as US dollars or the euro. You can change this <Base default> to any other

currency. When you load a conversion table, the data appears here initially.

The right pane, or Euro Table, records the fixed conversion rates used to convert values to or from

the national currencies of EMU countries by way of the euro. You can load this information into

the table from an external data source that provides euro rates, or enter it manually.

Base and Euro Currency Files

These two comma-delimited currency files contain conversion rates for the first quarter of 1999,

ready for loading into the Base Table and Euro Table, respectively. They include the three mandatory

columns (Date, Country Code, and Conversion Rate), as well as the optional Currency Label

column. You can change rates marked with a pencil icon. Initially, these have a default value of

1.0000. (After external rates are loaded, they cannot be manually updated).

The data used in these examples is for illustrative purposes only. It is not accurate. To obtain official

currency conversion rates for the euro, we recommend that you view the data available on the

European Commission Web site at http://europa.eu.int/euro

DATE,CURRENCY LABEL,COUNTRY CODE,CONVERSION RATE










EURO DATE,EURO CURRENCY LABEL,EURO COUNTRY CODE,EUROCONVERSION RATE



0,Austrian Schilling,AUT,13.7603

0,Belgian Franc,BEL,40.3399

0,Finnish Markka,FIN,5.94573

0,French Franc,FRA,6.55957

0,German Deutsche Mark,DEU,1.95583

0,Italian Lira,ITA,1936.27

0,Dutch Guilder,NLD,2.20371

0,Spanish Peseta,ESP,166.386

19990101,U.S. Dollar,USA,1.16741

19990201,U.S. Dollar,USA,1.17028

19990301,U.S. Dollar,USA,1.17330

Rates are expressed as the amount by which you multiply a unit of the given currency to buy a unit

of the base currency. In March 1999, 1 US dollar (base currency) cost 1.51 Canadian dollars, 0.60

British pounds, and 112.90 Japanese yen. 1 euro cost 6.56 French francs (fixed) or 1.17 US dollars

(sample rates, for illustration purposes only).

Conversions Requiring Euro Triangulation

During the transition period when EMU-member countries carry two legal tenders, special rules

apply for conversion to and from the national currencies. To be euro compliant, all rates must be

expressed in terms of the euro. Values must be converted from the original currency to the euro,

and then from the euro to the target currency, with rounding to the specified precision occurring

at each step. The required multiplication, division, and rounding operations are carried out auto-

matically in PowerPlay.

Use one or more data sources for the Base Table, with identical column names, to supply the con-

version rates for non-member countries, and for member countries prior to their EMU entry. Use

another data source, with different column names and a zero date, for your Euro Table. Or,

manually type in the fixed rates on entry into the EMU.

In Transformer, ensure that all source dates are sorted in ascending order, and that all date levels

are set to Unique. Remember that you can't change externally supplied rates, after they have been

imported into your Currency Table. Fixed rates are those not marked with a pencil icon .

Notes

● If you use euro triangulation, the date column in your data source must be sorted in ascending

order.

● You should clear the Timing options in the property sheets for all currency data sources so that

updates to the Currency Table do not interfere with category generation in the rest of your

model. If you decide to leave all Timing options at their default settings, ensure that the names

of the date columns for the base and euro currencies differ from the column name of your time

dimension, and from each other. Otherwise, Transformer can't differentiate between them.

● Check Model issues a warning if your model contains a Currency Table but none of the measures

have currency conversion enabled. You can ignore this message, but currency conversion will

not be available in PowerPlay.



● An EMU Entry Date is automatically specified for the euro (as the default currency in the Euro

Table). Conversion rates created for the <Base Default> currency before that date are set to

zero. They appear as "NA" in Transformer, and as zero in Model Definition Language (MDL).

Any use of zero as a conversion rate causes the data to which that rate applies to be treated as

missing in PowerPlay.

● The Currency Symbols and Decimal Formats that appear in PowerPlay are determined by the

Country Codes. You can override these defaults in the Currency Record property sheet.

Steps to Prepare Base and Euro Table Source Files

1. If you have not already done so, create Base Table source files and load them into the Base

Table.

See the first two steps in "Create a Base Currency Table" (p. 34).

2. Create one or more source files to populate the Euro Table portion of yourCurrency Conversion

Table.

Each of these files should contain

● a date column; for each EMU-member country, insert a zero value for the date; for countries

that are not EMU members, add one row for each time period in which conversion rates

apply, and insert the correct date

● a country code column, to identify the countries to which the conversion rates apply; if the

country already appears in the Base Table, the code must be identical to that used in the

Base Table

● the conversion rates, expressed as the amount by which you multiply a unit of the currency

to buy one euro

● optionally, labels for the currencies (if you don't specify a label, it is derived from the

country code)

Your file should appear similar to the second sample in the topic "Create a Euro Currency

Table" (p. 36).

3. Add a data source to your model for each source file.

Note: If you use more than one source file to populate the Euro Table, their column names in

your model must match. Ensure that column names specified for euro conversion differ from

those specified for the Base Table.

4. Double-click each source in the Data Sources list to open its property sheet, and click the

General tab.

5. Clear the Timing check boxes and the Sets the Current Period check box, or ensure that the

conversion date column has a different name from that of the time dimension. Click OK.

Steps to Create a Euro Table

1. From the File menu, click Currency Table and select the Use Euro Triangulation check box.



2. Confirm that the data source Associations window shows the correct Date, Country Code, and

Conversion Rate columns to populate the Euro Table. If not, select Modify in the pop-up menu

and make corrections.

3. Repeat Step 2 for the Currency Label column, if present. (This name may be derived from the

Country Code.)

4. For each country that is an EMU member and will appear in the Euro Table, perform the fol-

lowing steps:

● Right-click the country in the Base Table and click Properties to open the Currency Recordproperty sheet.

● Select the Member of the Economic and Monetary Union (EMU) check box.

● Accept the Default Entry Date (19990101) or type a different date, and then click OK.

Respond to the warning message about changing the date level by clicking Yes.

5. Click Load Table.

The Euro Table box is populated with the rates for EMU countries and any non-EMU <Base

Default> countries. Respond to the warning message about overwriting entries by clicking OK.

Note: The fixed rate on EMU entry (zero date) is marked by a special icon. Rates for the non-

EMU <Base Default> currency also appear.

6. Follow the steps in "Enable Currency Conversion" (p. 39).

Enable Currency ConversionSometimes your external data sources provide currency conversion rates at greater levels of detail

than are needed by your PowerPlay users. By default, Transformer applies conversion rates to the

lowest level of detail in the time dimension (typically, at the Month level). However you can change

this setting to a summary level, such as Quarter, in the Currency Record property sheet.

Steps

1. To specify the level in the time dimension to which the conversion rates apply, right-click each

currency in the Base Table or Euro Table in turn. Click Properties to open the Currency Record

property sheet and change the Date Level property to the correct level.

2. Click OK, or click Load Table again, to update the Currency Table data.

3. Open the property sheet for each measure that requires currency conversion, and click Allow

Currency Conversion (General tab).

Create or Update a Currency Table ManuallyIf your model design is simple and does not use external data sources to supply conversion rates,

you can manually create and update the Base and Euro Currency Tables.



If you used an external data source to create the initial Currency Table, but decide to maintain it

manually, remember to clear the Use an External Currency Data Source check box before proceeding.

Steps

1. Right-click in the Base Table box, or the Euro Table box if applicable, and click Add New

Currency.

2. In the Country Code box, select a code, set other properties as required, and then click OK.

3. Click the conversion rate that you want to change and make the required changes.

4. Repeat step 3 for every rate that you want to change.

5. Click OK.

Use Multiple Data Sources in Your Model: OverviewModels that use multiple data sources offer several advantages:

● Measure allocation. Transformer can take measures that are normally unrelated to some portion

of the dimension map and allocate their values to those portions. For example, sales-tracking

systems do not generally include information about staffing levels at each branch. However,

by including a data source that provides this information, you can derive valuable information,

such as the average sales per employee.

● Performance improvements. By splitting a single, large data source into smaller, partially nor-

malized sources, you can shorten the processing time if the columns in all measure-containing

sources are associated with unique levels in the model. This is because Transformer can associate

such measures directly, without regard to their context in the level hierarchy.

● Integration of data from various sources. Models with multiple data sources can combine data

from different departments or locations, even if they use different storage systems.

Notes

● Models based on multiple data sources are supported in all editions of PowerPlay and Trans-

former.

● Transformer does not perform database joins between like-named columns. Rather, it associates

the data values in each source with the categories that are generated from the columns. To

create source data that joins database tables, use IBM Cognos Series 7 Impromptu, IBM Cognos 8

Framework Manager, or some other data access application.

View the Scope of a Data SourceData sources may be associated with all, or a portion, of the dimension map. To make the proper

associations, all column names related to the same level in the dimension map must match exactly,

and the levels with which these source columns are associated must be either unique or fully qualified

by one or more higher levels in the hierarchy.



Additionally, scope depends on the relationship of the source columns to the levels in the dimension

map, as shown in the accompanying illustrations.

Level Derived Directly

The level takes its category values from a column in the source. For example, this source contains

columns that provide category values for all levels.

Level Derived Indirectly

The source is not related to the level, but is related to descendants of the level. For example, this

source contains a column that provides category values for the unique level Product No., but no

columns for the ancestor levels. Product Type and Product Line may be manual levels, or their

category values may come indirectly from columns in other data sources.

Level Derived from a Source with Missing Columns

The level cannot take its values from the source because the source lacks columns for ancestor

levels, and the level is not unique. The source does not contain enough data to map column values

to a level without knowing the entire context. For example, this source contains a column that

provides category values for the level Product No., but not for the ancestor levels. The context of

the Product No. categories cannot be ascertained because the level hasn't been declared unique.

Example

The following sources are used for a simple sales analysis model:



The Go_sales source provides category values for levels in the Dates, Products, and Regions

dimensions respectively. It also contains values for the Revenue, Cost, and Order Qty measures.

The Forecast source contains planned revenue for each Product Line. Its YEAR and PROD_LINE

columns are renamed to match their counterparts (Order Date and Product Line) in the go_sales

source:

Because go_sales contains columns for all three dimensions, its scope covers all the dimensions:

However, the Forecast source contains columns associated with the highest levels of the Order Date

and Product Line dimensions, but no columns for the descendant levels. Its scope is limited to Year

and Product Line levels. Category values for lower levels in these dimensions come indirectly from

the go_sales source:

Tips

● You can manipulate a scope dimension map in the same way as a regular dimension map. The

manipulations affect all open dimension maps.

● To see the scope, you can also use the pop-up menu in the dimension map or the Data Sources

list.



● You can change the default colors of the scope dimension map in the Dimension Map tab

(Preferences dialog box).

Steps

1. In the Data Sources list, select the source whose scope you want to see.

2. From the Edit menu, click Show Scope.

Control When Transformer Reads the Source DataIn models with multiple data sources, you can control whether a source is read, and when: during

category generation, during cube creation, or both. Use this capability to avoid unnecessary pro-

cessing and to improve model efficiency. Three typical scenarios are as follows:

● After Transformer has read your purely structural data sources, and populated your model

with data for the required category structure, you can change the timing so that these static

categories are not regenerated every time you create the PowerCube.

● For your volatile structural data sources, set the timing to update categories every time you

create the PowerCube.

● For transactional data sources, where the measure values are constantly changing, select the

PowerCube Creation check box for the data source, and then the Default option below it.

Steps

1. Open the Data Source property sheet for the source file for which you want to control the

timing.

2. Click the General tab.

3. In the Timing box, select the Generate Categories check box to query this source whenever

categories are generated.

4. Still in the Timing box, select the PowerCube Creation check box, and then click the appropriate

option to specify what occurs when cubes are created or updated:

● Default. Transformer reads any columns in the source that relate to levels in the model, to

see if they are associated with measures. If the source is purely structural, the cube is not

created or updated, and an error message is issued. Use this option in a production envir-

onment, for transactional data sources and structural data sources that contain non-static

data.

● Generate Categories Only. Transformer only queries for structural information when the

Create and Update Cubes command is invoked, but does not include measures from the

source (if they exist).

● Create the PowerCubes. Transformer queries the source and creates or updates the cubes,

even if the source lacks measures. Use this option in a design and development environment,

to override any error messages associated with the Default option.



5. Click OK.

Use Local Data Sources: OverviewTransformer supports a wide range of local data sources:

● delimited ASCII (.asc) and comma-separated variable (.csv) files

● Impromptu Query Definition (.iqd) files, which can query local or server-based databases

● fixed-field text files

● local databases, including Microsoft Access, which can specify SQL queries against local or

server-based databases, Clipper, dBase, FoxPro, and Paradox

● PowerHouse portable subfiles

● spreadsheet crosstabs and databases, including Excel and Lotus 1-2-3

Importing data is a two-stage process. First, you create the source files that Transformer will use,

and then you add them to the Data Sources list. Models can use an unlimited number of sources

or columns. However, Transformer does not perform joins between data files. Your sources must

contain sufficient data to provide the necessary context for each drill-down path in the model.

Notes

● For most local data sources, you can specify the character set used (DOS Code Page or Windows

ANSI).

● Although Impromptu Query Definitions and Microsoft Access files are local sources, they can

provide access to server-based data. For very large source files, .iqd files may provide better

performance than Microsoft Access files when generating categories or creating cubes (more

efficient passage of queries through the ODBC layer).

Use Spreadsheet Data SourcesYou can use data from Lotus 1-2-3 or Microsoft Excel spreadsheets as source files, in either crosstab

or database format.

Before using a crosstab as a source, you must define one or more named ranges in the spreadsheet.

These ranges determine which data will become columns in the model. When using a database as

a source, you must define a named database range in the spreadsheet, and also specify that range

in Transformer.

Transformer reads the named ranges from the crosstab or database, and the Data Class (text,

numeric, or date) for each cell value. For more information about how to define named ranges, see

the documentation provided with your spreadsheet software.

Example 1

This example shows one page of a multipage crosstab.

EDCBA



Q4Q3Q2Q1Glassware Division1

2

$20,000$15,000$10,000$10,000Income - Net Sales3

$2,500$2,500$2,000$2,000Expenses - Salary4

$800$800$800$800Expenses - Rent5

$600$600$700$600Expenses - Advertising6

$4,000$2,000$1,500$1,500Expenses - Production7

$12,100$9,100$5,000$5,100NET INCOME8

For this crosstab, the following ranges have been named (range names automatically become the

columns in a Transformer model).

RangeNameRangeName

A4 .. A7ExpenseA1Division

B4 .. E7Expense AmountB1 .. E1Quarter

B8 .. E8Net IncomeB3 .. E3Income

For a model requiring only the DIVISION, QUARTER, EXPENSE, and EXPENSE_AMT columns,

Transformer reads all the expense types and expense amounts for each quarter and for each division.

AMTExpenseQtrDivision

$2,000Expenses - SalaryQ1Glassware




...

$1,500ProductionQ1Glassware

$1,500ProductionQ2Glassware

The Data Sources list shows the following:



Example 2

This example shows a simple spreadsheet database.

A B C D

1 EMP_NO NAME DEPT SALARY

2 0256 Wilson, K. TECHW $50,000

3 0141 Barnes, J. DESIGN $60,000

4 0724 Paul, N. DESIGN $70,000

5 1290 Power, R. DESIGN $80,000

The database has a named range.

RangeName

A1 .. D5DataTable

Transformer reads rows 1 to 5 in the range DataTable. The Data Sources list shows the following:

Notes

● Range names from Lotus 1-2-3 that contain spaces are not valid. Replace the space with an

underscore; for example, change a Lotus 1-2-3 range named "NET INCOME" to

"NET_INCOME."

● Before generating any categories, Transformer verifies that the named range represented by

each column in the model still exists in the spreadsheet, and that the spreadsheet still contains

the same number of pages, rows, and columns as when you first identified it as a data source.

If Transformer detects a difference, it issues a warning, and prompts you to use the Modify

Columns command to add, modify, match, or delete columns.

● If the Data Sources list contains columns identifying named ranges that you don't need, you

can delete them from the list without affecting the spreadsheet.

Steps

1. Right-click the Data Sources list and select Insert Data Source to open the New Data Source

wizard.

2. In the Data Source Name box, type a name for the new source.

3. In the Source Type box, select one of the following:



● Lotus 1-2-3 crosstab

● Excel crosstab

● Lotus 1-2-3 database

● Excel database

4. In the Local Data File box, type the file name of the spreadsheet file.

5. For a Lotus 1-2-3 or Excel database, select a range name from the Table or Range box.

6. When you have specified all of the required parameters, click Finish.

Use PowerHouse Data SourcesYou can use a PowerHouse portable subfile as a data source for the model.

This table shows how Transformer interprets the components of a PowerHouse portable subfile

dictionary.

Transformer modelPortable subfile dictionary

Column Original NameElement Name

Column Name (The PDL multiline character (^) is replaced

by a space character.)

Element Heading

Item Datatype:

Data Class Text, and Size- Character

Data Class Unspecified (default); Measure Type 32-bit

Integer

- Portable Zoned

Data Class Unspecified (default); Measure Type 64-Bit

Floating Point

- Portable Float

Data Class Date; Date Input Format Predefined- Portable Date

Notes

● Transformer ignores the PowerHouse dictionary attributes Output Scale and Decimals. Instead,

columns generated from a portable subfile have the Output Scale and Decimal Places attributes

initially set to zero. You can modify these attributes by using the Format tab of the Measure

property sheet.

● Transformer presents integer and float data retrieved from PowerHouse portable subfiles

according to .psd datatype. Integers stored as Portable Zoned in the subfile appear as signed,

zero-filled numbers in Transformer. For example, the number 16, stored in Portable Zoned,



Size 6 format, appears as +00016. Float data stored as Portable Float in the subfile appears in

scientific notation. For example, the number 812,333.65, stored in Portable Float format,

appears as +.8123336500000000000E+06.

● You can specify either the portable subfile dictionary (.psd) file or the data (.ps) file when

pointing to a PowerHouse portable subfile.

Steps


wizard.


3. From the Source Type box, select PowerHouse Portable Subfile.

4. In the Local Data File box, type the file name of the PowerHouse portable subfile.

5. Accept the default Character Set or specify an alternative, then click Finish.

Use Local Databases as Data SourcesTransformer uses the Microsoft ActiveX Data Objects (ADO) driver to access data from any of the

following databases:

● dBase

● Microsoft Access

● Paradox

● FoxPro

● Clipper

Notes

● You can use an Access Query that points to an ODBC data source as a means of getting server-

based data into Transformer.

● If you use an Access Query as a source, the columns that appear in the Data Sources list are

the result of the SQL query run against the database.

● If you use Paradox tables with a sorting key other than ASCII (the default setting in the Jet

engine for Paradox), before you import the tables you must include the .px file in the same

location as the .db file, and you must change the following Windows Registry key to a value

corresponding to the sorting key used when creating the table:

$HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Jet\3.5\Engines\Paradox\CollatingSequence



Valid values are ASCII, International, Norwegian-Danish, Japanese, and Swedish-Finnish.

Steps


wizard.


3. From the Source Type box, select one of

● dBase Table

● Access Table

● Access Query

● Paradox Table

● FoxPro Table

● Clipper Table

4. In the Local Data File box, type the name of the database table or query source file.

5. Select a Table or Range and a Character Set, as appropriate.

6. When you have specified all of the required parameters, click Finish.

Use Text Data SourcesTransformer can use plain text (ASCII) files stored in the following formats:

● Delimited-Field Text

● Delimited-Field Text with Column Titles

● Fixed-Field Text

● Fixed-Field and Record without CR LF

You should not rely on the file extension as an indicator of file type. Text files are often saved with

a .txt file extension, or .asc or .csv if delimited-field, but they can also be saved with any other file

extension, or no extension at all.

If you use fixed-field text files you must define the columns manually, because Transformer has no

information about how columns in the file are defined in the source file.

Steps


wizard.


3. From the Source Type box, select one of the following:



● Delimited-Field Text

● Delimited-Field Text with Column Titles

● Fixed-Field Text

● Fixed-Field and Record without CR LF

4. In the Local Data File box, type the file name of the text file.

5. Accept the default Character Set or specify an alternative, enter any other required parameters,

then click Finish.

Use .iqd Files to Access Server DataIf IBM Cognos Series 7 Impromptu is installed on your computer, you can use it to create Impromptu

Query Definition (.iqd) files that Transformer can use to query both local and server-based databases.

You can also use .iqd files with model queries defined in Framework Manager, the modeling tool

used with IBM Cognos 8 Business Intelligence components.

In Series 7, you can run Impromptu directly from Transformer to create new data sources by

clicking the Impromptu button on the toolbar. After you have used Impromptu to create an .iqd

file, you can query directly from Transformer. For an up-to-date list of the relational and non-

relational databases that Impromptu supports, visit IBM Cognos on the Web (http://www-01.ibm.

com/software/data/cognos/).

If you have the Series 7 Transformer UNIX edition installed, you can use .iqd files from UNIX

servers. The server connection information on the client computer must match that on the server,

or else Transformer on UNIX will not be able to read the database. Transformer embeds .iqd files

that are used as data sources in the model. As a result, Transformer on UNIX can query the server

database when it receives model information from Transformer on Windows.

Example

Some of the data required for a sales-tracking model is stored in an Oracle database on a UNIX

server. Using Series 7 Impromptu, you query the ORDERS and ORDER_DETAILS tables, and save

the Impromptu report as an .iqd file. Transformer can then use the information in the file to access

data from the Oracle database. After you add the .iqd file to the Data Sources list, Transformer

accesses data directly from the server when generating categories and creating cubes.

Steps

1. Create the report you want in Impromptu, and save it as an Impromptu Query Definition (.iqd)

file.

Note: Ensure that the .iqd file is not in crosstab format.

2. In Transformer, right-click the Data Sources list and select Insert Data Source to open the New

Data Source wizard.


4. From the Source Type box, select Impromptu Query Definition.



http://www-01.ibm.com/software/data/cognos/


5. In the Local Data File box, type the file name of the .iqd file.

6. Accept the default Isolation Level or specify an alternative, then click Finish.

Define Columns: OverviewModel columns must remain synchronized with their associated data sources. You create or delete

columns, and examine or modify column properties, in the Data Sources list.

Transformer uses source columns to build dimensions and measures in the model. A column in the

model represents a column in the data source. The data column contains text, date, or code values

that become the categories in your model, or numeric values that you select as measures (performance

indicators) for your model. A data column can also contain values that will be used as alternate

labels, short names, or textual descriptions for categories.

For most of the supported source types, Transformer can automatically identify columns in the

data source and assign default column names and properties. You can change the assignments later,

if you like. If you create columns manually (as you must when using a fixed-field text file), you

need to identify only those data columns that are used in the model.

Naming Columns

When an identically-named column appears in two or more data sources, Transformer can associate

each column with a level that has its source column set to the common name. Transformer does

not use matching columns to perform joins on the source files. Instead, it uses the common name

to associate both columns with categories in the dimension map or with a measure.

For example, suppose that you add an Order Header source and an Order Detail source, each

containing a column called ORDER_NUMBER. Transformer cannot join an Order Header record

with an Order Detail record to create a composite record describing a sales line in an invoice.

However, Transformer can associate the values from both of these sources with a specific level in

one of the dimensions in the model.

To create and use a join between two columns, use a database query tool such as IBM Cognos

Series 7 Impromptu or IBM Cognos 8 Framework Manager, before bringing the source data into

the model.

Dates

To obtain date values for the categories in a time dimension, at least one data source must contain

date values. You can

● set the absolute range of dates allowed in the model

● match the date format in the model to the format in the source file

● specify whether the measure values associated with these dates apply to some or all of the levels

in the time dimension; see "Set the Degree of Detail for Dates" (p. 54).

In a model with multiple data sources, date columns may appear in several source files, and each

column may relate to the same time dimension. For example, a sales-analysis model might contain

one source file with order information (including the order date) and another with sales forecasts



by time period. For Transformer to relate values from both of these sources to the time dimension,

the columns that contain the date values in each source file must have the same name.

Calculated Columns

If you need more structural or numeric information in your model than is available from your

source data, you may be able to add it by using calculated columns. See "Define a Calculated

Column" (p. 61).

Note: For all data types except fixed-field text, when you add a source file Transformer automatically

creates the column definitions. For fixed-field text sources, you must describe the position and

length, in bytes, of each column in the source file.

Define Columns in Fixed-Field Text Data SourcesWhen your data source is a fixed-field text file, you must define the columns by using the Column

property sheet, because Transformer has no information about how columns in the file are defined.

For example, in the following file, each record contains 38 bytes, representing five fields:

12345678901234567890123456789012345678 <- Byte

19990103Dishwashers White 02140330

19990103Dishwashers Chrome 02870335

19990103Stoves White 04560508

19990103Stoves Beige 03110388

The first field is an 8-byte numeric field giving the date of a transaction. The next 14-byte field

indicates the product type, such as Dishwashers. An 8-byte field indicates the color of each item

sold. The last 8 bytes store values for the measure fields Cost (4 bytes) and Revenue (4 bytes). The

first record, therefore, represents a $330 transaction that occurred on January 3, 1999, involving

a white dishwasher.

If the source file is a fixed-field text type, you cannot use the Modify Columns command (Tools

menu) to match the source to columns in the event that the structure of the source file changes.

You must manually adjust the starting byte and width of each column on its property sheet.

You can define overlapping columns or define a column that includes several other columns.

Steps

1. Right-click the Data Sources list, select Insert Data Source, and use the New Data Source wizard

to add a fixed-field text source.

2. From the Edit menu, click Insert Column.

3. In the Column Name box, type a name for the new column.

4. In the Position box, type the starting position of the column in a record.

The first byte in a record is byte number 1.

5. In the Size box, type the width of the column in bytes.

6. Set other properties for the column as required, and click OK.

7. Repeat steps 2 to 6 for each column in the source file.



Scale Input ValuesSometimes, you may want to change the number of decimal places or significant digits in your

measures: to better reflect what your PowerPlay users are expecting in their reports, for example,

or to make calculations. Decimal values are read into the model based on a specified scale, defined

as the source value multiplied by 10 raised to the power of the input scale. This scaling method

allows Transformer to handle values without integer portions, such as .0003.

To properly scale your input values, set the

● Input Scale property for the source column

● Output Scale property for the measure that uses the source column

● Decimal Places property for the measure that uses the source column

Example

A source file in comma-separated variable (.csv) format contains the following values:

DATE,CUSTOMER,PRODUCT_LINE,PRODUCT_TYPE,PRODUCT,QUANTITY,SALE_AMOUNT

19990603,GO Outlet Montreal,Outdoor Products,Tents,40100,5,600.55



To convert these decimal values to the appropriate integer values, without changing the results in

the report, set the

● Input Scale for the column SALE_AMOUNT to 2; each value in the source is multiplied by 10

to the power of 2 (the input scale), to produce an integer value (600.55 becomes 60055, for

example)

● Output Scale for the measure to match the Input Scale for the SALE_AMOUNT column; the

measure value will be divided by 10 to the power of 2 (the output scale) when PowerPlay is

run (the scaled value 60055 will revert to 600.55, for example)

● Decimal Places property for the measure to 2, so that the correct number of decimal places

appears in the report (600.55 in our example)

Rounding does not occur when you scale numbers with Input Scale, Output Scale, and Decimal

Places. For example, in scaling $1,792,485.86 to $X.X million, using an Input Scale setting of 0,

an Output Scale setting of 6, and a Decimal Places setting of 1, yields $1.7 million, not $1.8 million.

Note: In Transformer, regardless of the precision supported by the source database, the overflow

limit for the 64-bit floating point storage type is 18 significant digits, excluding the decimal separator.

If your measure values have more significant digits after the decimal place than is specified in the

precision attribute of the model, the number is truncated and the last bit rounds up, rather than

down.

Steps

1. Open the Column property sheet for the measure which you want to scale.

2. In the Input Scale box, type a value (the number of decimal places in the source data, for

example), and click OK.



3. Open the Measure property sheet.

4. In the Output Scale box, to maintain consistency between the source data and the measures in

your reports, type a value that matches the Input Scale specified in step 2.

Note: The Input Scale doesn't have to match the number of decimal places in the source data,

nor does the Output Scale have to match the Input Scale. Enter values that make sense for your

users and reporting needs.

5. In the Decimal Places box, type a value for the number of decimal places to appear in the report,

and click OK.

Set the Degree of Detail for DatesOften, values for certain measures in a model apply to time periods that are not at the lowest level

in the time dimension. You can specify the actual degree of detail to which the date value applies

in the model.

The Degree of Detail that you set for a column must be supported by the date values stored in the

associated column in your data source. For example, you cannot specify Day for the Degree of

Detail if the date values are stored as YM only (for example, 199902).

If a particular measure has meaning only to a certain level in the time dimension, you can allocate

values for that measure to lower levels. See "Allocate Measures: Overview" (p. 115).

Example 1

In sales analysis systems, actual sales are typically derived from the invoice information that accu-

mulates as orders are placed and filled. Each sales transaction has the date on which the order was

filled. In contrast, sales forecasts are projected for months or quarters, not for days. Similarly,

budgets for sales organizations are usually prepared by the month or by the quarter. You can use

the Degree of Detail property to tell Transformer the level of detail to which forecasts and budgets

apply in the time dimension.

Example 2

Sales figures are stored in the following format in the source:

WEEK,CUSTOMER,SALES_REP,TOTAL_SALES

19990208,Fresh Air Lte 4,Francoise LeBlanc,4977.99

19990215,Fresh Air Lte 4,Francoise LeBlanc,2955.85

Although the date values are specified in YMD format, the associated measure values are actually

weekly sales summaries by sales representative. You can use the Degree of Detail property to tell

Transformer to treat the sales figures as weekly rather than daily values.

Steps

1. Open the property sheet for the relevant date column, and click the Time tab.

2. In the Degree of Detail box, select the date level appropriate to the measure in the source.

For example, if the source contains a measure that provides monthly forecast values, select

Month.



Use Monthly or Quarterly Time ArraysIf your transactional data is stored as quarterly or monthly values, but you generally roll up this

information into yearly results, it may be more efficient to define the columns in your model as

members of a time array, rather than as individual measures. A time array consists of four or twelve

adjacent columns that contain quarterly or monthly values for one year.

The data source or sources for the time array must contain at least one date column, in addition to

the columns that represent the individual elements of the array. The value in the date column must

be the same for all data in the array.

To ensure that Transformer processes the array correctly, use the first month of your fiscal year as

the date in your date column. This specifies the year-begin date for the year in which the array

applies.

You can define more than one array per model, either in one source file with all arrays the same

type (quarterly or monthly), or in multiple source files, with one array type for each file (for example,

a different array for each year of data). Use the same format in all date columns, and a different

name for the first month or quarter in each array.

When you create the model, clear the Run AutoDesign check box. Otherwise, all the measures

appear in the Measures list before you have defined the array, and you must delete them before

beginning the array definition.

An array is treated as one object. If you delete any column that is a member of an array, Transformer

automatically deletes all other member columns in the same array.

Example 1: Consolidated Quarterly Data for a Single-Source Model

In this example, the source data contains the total number of sales for each product by quarter.

You can define the Q1, Q2, Q3, and Q4 columns as members of a time array.

DATE PRODUCT Q1 Q2 Q3 Q4199901 Product1 100 200 150 400199901 Product2 100 175 150 350199901 Product3 75 100 100 100200001 Product1 110 210 160 420200001 Product2 125 200 175 375200001 Product3 125 150 150 150

The following results will appear in your OLAP report.

1999+200020001999

1750900850Product1

1650875775Product2

950575375Product3

435023502000TOTALS



Example 2: Monthly Data for a Multiple-Source Model

In this example, there is one data source for each year of data. Each contains the total number of

items sold by retailer type (independents versus department stores), by month. You can define the

Month01 column in each source as the first column in a twelve-member array. Rename the initial

column names so Transformer can differentiate between the two; for example, you can change the

initial column in the 2000 array to A2Month_01, to distinguish it from Month_01 in 1999.

The consolidated data for Array 1 (1999 Months) is as follows:

PERIOD RETAIL TYPE 01 02 03 04 05.06.07 08 09 10 11 12

199901 Independent 05 06 07 05 04 03 06 04 08 02 01 09

199901 Department 10 13 07 05 14 15 06 15 08 12 10 05

The consolidated data for Array 2 (2000 Months) is as follows:

PERIOD RETAIL TYPE 01 02 03 04 05.06.07 08 0910 11 12

200001 Independent 10 13 07 05 14 15 06 15 08 12 10 05

200001 Department 20 26 14 10 28 30 12 30 16 24 20 10

In your PowerPlay OLAP report, you can nested charts to show how many items were sold by each

retailer type for the 12-month periods beginning 1999/01 and 2000/01. You can also show the

totals sold by all retailers for each year, and for all years.

1999+200020001999

18012060Independent

360240120Department

540360180TOTALS

Steps

1. Before importing any data into Transformer, ensure that your source file or files contain

● groups of contiguous columns (four columns for a quarterly array, 12 columns for a monthly

array)

● a date column with the same value for all members in the array (generally the first month

in your fiscal year)

2. Use the New Model wizard to add the source file or files to your model. Clear the Run

AutoDesign check box or delete all objects in the Measures list, so you can begin the array

specification process with a blank slate.

3. Double-click the first column that you want in the time array, to open its property sheet (for

example, Month_01).

4. Click the Array tab.

5. In the Array Type box, click 4 Quarters or 12 Months.

Transformer automatically changes theData Class of the subsequent columns toArrayMember.



6. In the Date Column box, click the column containing the starting month of your fiscal year.

This is the date column (generally the same value as the first month in the array, in YYYYMM

format).

7. In the Start Month box, type the month in which the fiscal year begins (default: 1).

8. Click OK, and drag the column to the Measures list.

9. To add another data source, click the Data Sources list, click the Insert icon, and repeat steps

2 and 3.

10. In the Column Name box, enter a different name if required to distinguish this column from

the first month or quarter in the previous array. Optionally, enter additional explanatory notes

in the Description tab.

11. Repeat steps 4 to 8 to add this new array to your model.

12. After you specify all the arrays you want in your model and ensure that theMeasures list contains

the initial column of each array, click Check Model (Tools menu).

13. If there are no problems with your design, create the cube and confirm your results in PowerPlay.

Tip: Use the Rename Label command (Edit menu) to provide more meaningful headings for

your array columns.

Specify Date Categories in Two CenturiesIf your source data represents date values using two-digit years, you must ensure that your data is

incorporated correctly into your model. By default, Transformer interprets the years 00 to 19 as

2000 to 2019, and the years 20 to 99 as 1920 to 1999.

Example

You know that your source data contains date information in YYMMDD format for the years

1980 to 2020. When you generate categories for your model, you see that your date categories stop

at the year 2019.You want to modify the way the categories are generated so that the dates extend

to 2020, as reflected in your source data.

To solve this problem, open your Trnsfrmr.ini file and reset the CenturyBreak option to the last

two digits of the first year that you want to appear in the earlier century (80, in this example). Years

80 or higher will be generated in the 20th century (1980-1999). Years less than 80 will be generated

in the 21st century (2000-2079).

Tip: To show an overlapping span of date ranges in both centuries (1900 to 1999 and 2000 to

2020, for example), you must convert your source data to the four-digit (YYYY) date format.

Steps

1. Open your Trnsfrmr.ini file.

2. Find the CenturyBreak entry. If you can't find the entry, then it is set to the default.

By default, Transformer interprets the years 00 to 19 as 2000 to 2019, and the years 20 to 99

as 1920 to 1999.



3. Modify or create the CenturyBreak entry to match the date values in your data source. Type

the following:

CenturyBreak=xx

where xx is the last two digits of the first year that you want to have appear in the earlier century.

4. Save the Trnsfrmr.ini file.

5. Close and restart Transformer so that your changes take effect.



Chapter 3: Map Your Data into Dimensions

A dimension is a hierarchical structure that represents a major segment of your business information.

For example, in a PowerPlay model for sales analysis, you might include these dimensions:

● Dates of Sales (Time)

● Locations of Sales (Region)

● Product and Purchase Details (Products)

● Customer Information (Customers)

Each dimension consists of one or more drill-down paths that typically contain several drill-down

levels. The time dimension may consist of the years, quarters, months, weeks, and days when sales

were made. The Locations of Sales dimension may be grouped into regions, states, and cities. The

Products dimension may organize your sale items by type, brand, model, color, and packaging. By

organizing your data into hierarchies, you can analyze results at various levels of detail.

Where Dimensions Appear

In Transformer, dimensions appear in the dimension map and the dimensions pane of the diagram.

From either of these two locations, you can manipulate the dimensions or their component categories

in various ways, or open the relevant property sheets.

In PowerPlay, the cube dimensions appear on the dimension line and in the dimension viewer. By

making selections from either of these locations, users can explore the cube and analyze its measures.

Alternate Drill-Down Paths

Some model designs allow users to drill down to data by more than one path. The dimension has

a primary drill-down path and one or more alternate drill-down paths, each of which contains one

or more levels. For example, a Regions dimension allows drilling by two distinct paths: Region,

State, and City, or Branch and City. These alternate drill-down paths converge at the City level.

Process

You can create dimensions by using AutoDesign, in which case the dimensions are structured for

you based on patterns and relationships in the source data. Alternatively, you can drag columns

from the Data Sources list to the dimension map.


Make a Preliminary Design with AutoDesignAutoDesign helps you make preliminary designs and runs whenever you create a new model, unless

you clear the Run AutoDesign check box in the New Model wizard. AutoDesign analyzes the data

types, column names, and structural framework of the source file, and places

● dates in the time dimension (if the dates are not in a predefined format, Transformer will prompt

you)

● columns with numerical values in the Measures list

● all remaining columns on the dimension map, by using a best-fit approach

You will probably need to change the dimensions that AutoDesign creates. For example, AutoDesign

does not distinguish numeric data that represent quantities from numeric codes, such as order

numbers. For a more accurate initial model, set the Data Class for each column and specify which

source columns are measures by dragging those columns from the Data Sources list to the Measures

list.

If your data source is a fixed-field text file, identify the columns in the query before you run

AutoDesign. See "Define Columns in Fixed-Field Text Data Sources" (p. 52).

You generally use AutoDesign to generate the dimensions, source levels, and measures of a new

model. However, you can also run AutoDesign on an existing model, to add dimensions and source

levels from as yet unused columns.

Steps

1. Add a source file to the model. See "Create a New Model" (p. 30)

2. If you did not check the Run AutoDesign check box in the New Model wizard, from the Tools

menu, click AutoDesign.

Create a New DimensionYou can create a dimension for each aspect of your business that you want to analyze in PowerPlay.

Steps

1. In the dimension map, click on the dimension line to ensure that no level is currently selected.


3. In the Dimension Name box, type a name for the new dimension.

4. In the Dimension Type box, click

● Regular, to create any dimension that does not track time

● Time, to create a dimension that contains periods such as years, quarters, months, and

days

5. Click OK.



For a Time dimension, specify whether to create the standard time levels (year, quarter, and

month): select the column that contains the date values from the Associations box, and click

OK.

Define a Calculated ColumnA calculated column is an expression that uses other columns, functions, and constants to derive

new data for the model. Calculated columns can be used in a dimension structure

● to create exception dimensions, or new ways of slicing and dicing your data

● to create customized date values

● to produce new measure values

Calculated columns provide a way to set up exception dimensions based on a calculation from

existing source columns. For example, you could use existing REVENUE and COST columns to

create a MARGIN_RANGE column that offers new insights into the data, as seen in PowerPlay.

To obtain margin range values such as Low, Medium, and High, you calculate Gross Margin by

using the formula (Revenue - Cost) / Revenue and insert the formula into the If-Then-Else statement:

if("Gross Margin" < 0.50) then ('Low') else (if ("GrossMargin" > 0.70) then ('High') else ('Medium'))...

You can also add a calculated column that consists of numeric data to be used in measure allocation.

Example

You have two sources of data about a professional sports team. The first source contains data for

individual players, and the second contains salary figures for each team.

You add a calculated column representing each team's forecast salary based on a 13% increase for

the next year, such as

"CURRENT_SALARY" * 1.13

If you drag the calculated column to the Measures list, you make these figures available for allocation

to the player level, from the other data source. In PowerPlay, your users can see each player's

forecast salary.

Notes

● When a calculated column is used as a measure, the value is always derived before any rollup

takes place. This is effectively the same as using a calculated measure. However, when the

timing of a calculated measure is set to Before Rollup, consolidation is not done. So, to

optimize both cube size and PowerPlay performance, use calculated columns rather than before-

rollup calculated measures. For information on consolidation, see "Consolidate Data: Overview

" (p. 120).



● The list of functions available in the expression editor varies with the Data Class specified. For

example, the First-of-Month and Today functions are available only if the Data Class is set to

Date.

● If your users request currency conversion, the column is initially calculated by using the default

(base) currency; after that, conversion occurs dynamically in PowerPlay, followed by rollup.

Steps

1. Click the Data Sources list to make it active.

2. From the Edit menu, click Insert Column.

3. In the Name box, enter a name for the new column.

4. In the Type box, click Calculated, and then click the Calculation button.

If you haven't specified a Data Class, Transformer prompts you to specify whether the column

consists of text, date, or numeric data.

5. In the left pane of the expression editor, expand the Columns, Functions, and Value folders as

needed, select each parameter you want to use, and click the arrow to insert it into the calculation

expression (right pane of the editor).

6. When the expression is complete, click OK.

Add Levels to a Dimension: OverviewEach of the dimensions in a model contains one or more levels that represent the information hier-

archies that may be explored in PowerPlay. Each level appears in the dimension viewer of PowerPlay

as a drill-down layer. You can adjust or rename the levels to suit the organization of your data

from the category viewer in Transformer.

A model may have the following level types:

● Source levels contain categories that are generated from or matched to column values in the

source data. Each source level is associated with one or more columns in the source file.

● Manual levels are drill-down layers not associated with source columns. You create and

maintain the categories in a manual level, and the manual level name in the dimension map is

preceded by a special icon .

Process

You can add levels to a dimension by using AutoDesign, by dragging source columns onto the

dimension map or the levels section of a dimension diagram, or by using the Insert command in

either the dimension map or the levels section of a dimension diagram, and then specifying an

Association for the new level.

You create a manual level in much the same way, but you avoid specifying a source column for the

level.



Add Source Levels to a DimensionSource levels obtain values for their categories from columns in the Data Sources list. You can add

source levels to a dimension either on the dimension map or in a dimension diagram.

Example

Your initial model has the following levels:

You add sales representatives to the Region dimension by selecting the Sales Rep column from the

Data Sources list and dragging it to the indicated spot:

Steps

1. In the Data Sources list, select the column for which you want to create a source level.

2. Drag the selected column to the appropriate location on either the dimension map or an existing

level in the dimension diagram.

On the dimension map, Transformer shows a small highlighted box where it will create the

new level.

On the dimension diagram, Transformer inserts the new level to the left of the level on which

you drop the source column.

3. Optionally, open the Level property sheet to set or modify the level properties, and then click

OK.

4. To add a new source level, select an existing level on the dimension map or dimension diagram

and click Insert Level from the Edit menu. Specify the Association for the new level.

5. To create a default date dimension with Year, Quarter and Month levels, set the Data Class

(General tab, Column property sheet) to Date, and then drag the column to the dimension map.

Add Manual Levels to a DimensionManual levels provide a means of grouping categories from various source levels under a new,

special category, or allow for intermediate groupings where there are too many child categories to

be easily seen in PowerPlay.

Because categories in manual levels typically connect to source categories in a lower level, new

source categories that are not linked to a parent manual category may appear during the generation

process. You can set up a temporary placeholder category in the manual level, where these new

categories can be placed by default. See "Set Up a Manual Level for Unknown Categories" (p. 99).



The manual level name in the dimension map is preceded by a special icon .

Notes

● You can use a manual level as an orphanage for any new categories that are subsequently

generated for the dimension.

● You can't use manual levels for subsets in PowerPlay.

● You can also create manual levels in the dimension map. Click Insert Level from the Edit menu

when a level is selected. Transformer recognizes the new level as a manual level if you leave

the Associations box empty.

● If you are building the same dimension in several models, create a spreadsheet to hold your

manual categories and their associated source categories. This spreadsheet can become the data

source for each model.

Example

Your staffing model contains a dimension with 14 categories at the country level:

You want to subdivide the Regions dimension so you can analyze your data more easily in PowerPlay.

You add a manual level that groups these countries by geographic region. You add the categories

Americas, Europe and Far East, and link USA along with its child categories to the Americas region.



You then link the remaining country categories to their appropriate regional categories.

Steps

1. Open the diagram for the dimension in which you want to create a manual level.

2. At the top of the diagram, position the pointer over the right side of an existing level.

The pointer changes to a cross-hair.

3. Drag the pointer to a position between two levels and release the button.

Transformer creates a new level and opens its property sheet.

4. In the Level Name box, type a name for the new manual level, and click OK.

5. For each intermediate category required in the manual level, create a category manually.

See "Create Categories Manually" (p. 67).

6. Connect the categories created in Step 5 to the appropriate child source categories in the next

lowest level.

Create an Alternate Drill-down PathBy setting up an alternate drill-down structure in a dimension, you can provide different perspectives

on the data in PowerPlay. Each such path connects to the primary path at the convergence level.

To correctly connect several parent categories to the same convergence category, each category in

a convergence level must be unique and unambiguous. In other words, no two categories in the



level can be derived from the same source value. Transformer prompts you to confirm uniqueness

when you create an alternate drill-down path.

Because the category values at the convergence level and below are shared by all drill-down paths,

removing or changing a category in one path at or below the convergence level immediately affects

the same category in all other drill-down paths.

Notes

● For more information about the uniqueness attribute, see "How Data Sources Relate to the

Dimension Map" (p. 234).

● When you create alternate drill-down structures in a time dimension, you must follow several

specific rules to select the convergence level and the properties of each drill-down path.

● You can also create an alternate drill-down path by dragging a column to the dimension map,

as depicted in the following diagram.

Example

The time dimension Order Date provides two distinct paths for arriving at the Month level:

● a path based on all years, including Year, Quarter, and Month

● another path based on the fiscal year, including Fiscal Year, Fiscal Quarter, and Month

In PowerPlay, users can drill down to the month level by either the calendar year or their fiscal

year.

Steps

1. In the dimension map, select the level that is to be the convergence level for the new alternate

drill-down path.

For a level to be a convergence level, it must be designated unique and the category values in

the level must have unique and unambiguous source values.

2. From the Edit menu, click Create Drill-Down.

Note: If the level you selected is not already defined as unique, Transformer asks whether there

can be two categories in this level with the same source value. You must click No to create a

new drill-down path.

3. Optionally, add intermediate levels to the new drill-down path.



Create Categories ManuallyManual categories may be added to either source levels or manual levels, in any dimension. To add

any necessary manual categories to your manual level, use the category viewer.

If you create a category where no level exists, such as in a dimension that does not yet contain any

levels, Transformer automatically creates a new level and opens the property sheet of the new cat-

egory.

If you create a category and then click Cancel on the Category property sheet, Transformer leaves

the category in place. You must delete the category if you no longer want it.

Example

Your staffing model contains a dimension with 14 categories at the country level:

If you add a manual level called Regions above the Country level, you can add the categories

Americas, Europe, and Far East in the Regions level. You can then link each category in the

Country level to the appropriate Regions category, beginning with the link between the USA cat-

egories and Americas.

You then link the remaining country categories to their appropriate regional categories.



Steps

1. Open the diagram for the dimension in which you want to create a category.

2. Expand the category viewer (right pane) as required to show the parent of the category to be

added.

3. Position the pointer over the right side of the parent category.


4. Drag the pointer to the right and release the button under the level that is to contain the category.

Transformer creates a new category and opens its property sheet.

5. In the Category Code box, type a name that uniquely identifies the category in the dimension.

6. If the category is in a source level, enter the name of the associated source column in the Source

Value box.

For the category in the model to be matched accurately, the category name must be unique

among all the source categories owned by the parent.

7. Set other properties for the category as required, and click OK.

Create Calculated CategoriesYou can use calculated categories to add commonly requested calculations to your model, such as

month-by-month growth or market share. Because the calculations are computed in Transformer

and then added to the cube, the results immediately appear in PowerPlay for every measure.

Category calculation is based on a formula and a set of categories to which the formula applies.

You compose the formula with the help of an expression editor and a selection list of functions and

operators. You then specify the categories to which the formula applies: either a single category, a

category set, or a level, all in the same dimension.

You can declare calculated categories at the dimension level (by using the Dimension property

sheet), or individually. When created at the dimension level, new calculated categories become siblings

of the categories from which they are derived. The formula varies according to the context of each

category, and the calculated categories (in the following example, percent growth for each quarter)

are identified in the category viewer by their own icon .



To declare calculated categories individually, or to change them after they are created, you use the

Show Category Calculation command (Diagram menu) or the Calculated Category property sheet.

The applicable formula can only reference another single category, not a category set or a level.

Separately defined categories are useful when you want to create special groupings, such as

'Japan'+'Hong Kong'. They are identified in the category viewer by a calculated category icon .

Example: Percent Growth

Your time dimension includes several special categories, among which are Current Month and Last

Month. You can use these categories to create a new calculated category, named Monthly Growth,

which calculates the percentage change between these two items by using the Percent Growth

function. In PowerPlay, users can see the monthly growth of any measure included in the cube.

Steps to Create Calculated Categories at the Dimension Level

1. Open the property sheet for the dimension that is to contain the calculated category and click

the Calculation tab.

2. Click the Add button.

3. Enter a name for your calculated category in the Label box (Dimension Calculation Definitiondialog box).

Optionally, select theGroup Calculated Category Together check box to list calculated categories

together in your reporting application.

4. Click the Calculation button.

5. In the left pane of the expression editor, expand the Functions folder, select the mathematical

function you want to use, and then click the arrow to insert it into the calculation expression

(right pane of the editor).

6. Expand the Levels folder and select a Level, Single Category, or Category Set for the expression,

as appropriate.

See "Define Category Sets for Calculated Categories" (p. 70) if you want to use category sets.

7. In the Categories Diagrammer, expand the levels in the diagram, so you can select and drag

the appropriate items into your expression calculation.

Note: You can select, copy, or type a valid expression.



8. Click OK in the editor to save your finished expression.

Steps to Create a Single Calculated Category

1. Create a manual category in the level where you want the calculated category.

See "Create Categories Manually" (p. 67).

2. From the Diagram menu, click Change to Calculated Category.

3. In the left pane of the expression editor, expand the Functions folder, select the mathematical

function you want to use, and then click the arrow to insert it into the calculation expression

(right pane of the editor).

4. In the Categories Diagrammer, expand the levels in the diagram so you can select and drag the

single category you want to use in the calculation expression.

5. Click the Close button in the Categories Diagrammer pane and click OK in the editor to save

your finished expression.

Define Category Sets for Calculated CategoriesWhen you create a calculated category, various functions allow a category set to be specified as a

parameter. Sets are a convenient way of grouping categories when you don't want to include the

entire level in your formula. A set can also contain categories from different levels.

Steps

1. If you have not already done so, generate categories for your model, then open the property

sheet for the dimension in which you want to define a set.

2. Click the Calculation tab and then click the Add button.

3. Enter a name for your calculated category in the Label box (Dimension Calculation Definitiondialog box).

Optionally, select theGroup Calculated Category Together check box to list calculated categories

together in your reporting application.

4. Click the Calculation button.

5. In the left pane of the expression editor, select the function that you want to apply to your set,

for example Share, and then click the arrow to insert it in the calculation expression (right pane

of the editor).

6. Double-click Category Set and observe that Set 1 appears in the left pane of the expression

editor.

7. Expand the levels in the Categories Diagrammer so you can select categories for the set.

8. Select and drag the categories in your set, one by one, to the drop location for Set 1.

Observe that the right pane mirrors the left.



9. Double-click Single Category or Category Set again, as appropriate for your calculation, and

then drag the required category or categories to the drop location in the right or left pane,

respectively.

10. When you have defined all of your required sets and have selected, copied, or typed the rest of

your calculated expression in the right pane, click OK.

11. Click Generate Categories and, if you previously closed the diagram, reopen it by clicking Show

Diagram.

12. Expand the levels to show your new calculated categories and verify that they appear correctly

in the diagram.

Order Categories Within LevelsBy default, categories appear in the category viewer in the order encountered during query processing.

When a new category is generated, Transformer places it at the end of the child category list for its

parent category. You can sort the categories alphabetically or numerically, in ascending or descending

order, based on values in the source column or another column.

You may order categories in levels by

● dragging the categories from one spot in the diagram to another, to change their order explicitly

● specifying an Order By column for the level that contains the categories and then modifying

the Order Value property for individual categories in that level

When you specify an Order By setting for a level, Transformer assigns a specific order value to each

category in the level. These values appear in the Order Value box on the Order By tab (Category

property sheet) and are automatically applied whenever new categories are added to the model.

Example

In the following dimension diagram, the categories in the Product Type level appear in the same

order as in the source file:

If you want them to appear alphabetically, you can select the Product Type column as the Order

By column and sort in ascending order. Transformer orders the categories in the level accordingly:



Steps

1. Open the property sheet for the level containing the categories to be ordered, and then click

the Order By tab.

2. If the level is a convergence level accessible from two or more drill-down paths, then, from the

Drilldown box, select the drill-down path in which the categories are to be ordered.

Note: If the level is only a member of one drill-down path, the Drilldown box is not shown.

3. From the Sort By Column box, select the column whose values will be used to determine the

sort order.

4. In the Sort Order box, click either Ascending or Descending.

5. In the Sort As box, click either Alphabetic or Numeric, to specify whether values are to be

interpreted as text or numbers during the sort.

6. Click OK.

Order Categories Using a Global Preference SettingYou can specify that all categories in the model use a particular sort order, rather than manually

setting the order-by preference for each category individually.

You can still specify a different OrderBy association for a given level/drill combination. However,

if you do not, the category label is used as the sort value, and ascending is always the order used.

Steps

1. Open the model and, from the File menu, click Model Properties.

2. On the General tab, select Use the Preference Setting in the Default Category Ordering box to

order categories globally, and click OK.

3. From the File menu, click Preferences, and then click the General tab.

4. Select the Order Categories by Default check box and click OK.

Tip: You can check for the entry OrderCategoriesByDefault=1 in the [PowerPlay Transformer]

section of the trnsfrmr.ini file. The default location for this file is the installation_location\bin folder;

OrderByCategoryLabeling is the equivalent setting in UNIX.



Create Unbalanced Hierarchies Within a DimensionYou can use subdimensions to provide different levels of detail for specific categories, also known

as unbalanced hierarchies. For example, some branch offices may report product sales down to the

item level, whereas others may report only to the product level. You can create a subdimension for

those branches that report to the item level.

Categories in a subdimension are independent of levels in other parts of the dimension. Changes

made to levels outside the subdimension do not affect categories in a subdimension, and vice versa.

Transformer may create an unbalanced hierarchy if there is an overlap between the structural and

transactional data of a multiple query model, but the source data does not supply all of the values.

In the dimension map, a level that contains a subdimension has an ellipsis next to its name.

Note: Depending on your PowerPlay edition, portions of this topic may not apply to you. For more



Example

Your time dimension contains the levels Year, Quarter, and Month. You don't need Month values

for the previous year. By using a subdimension, you retain the levels in the current year while

removing them from the previous year.

Steps

1. Open the dimension diagram, and select the category below which you want to create a subdi-

mension.

2. From the Diagram menu, click Create/Delete Subdimension.

Transformer shows a box enclosing the selected category and all categories below it.

3. Modify the categories in and outside the subdimension, as required.



Define a Scenario Dimension and a Default Opening LevelScenario dimensions are not recognized as a distinct type in Transformer. However, you can flag

any dimension other than the time dimension so that its data never rolls up to the root category,

or a designated parent category, which is hidden.

We also recommend that you set a default category for the cube to open at. You can thereby ensure

that cubes containing budget values or other scenario-like data do not display zeros, n/a, or

meaningless numbers when opened by report authors or consumers.

You can designate a category that is a child of the non-selectable upper-level category in your

scenario dimension, as your new default. You then use the Hide the Value option with the scenario

dimension root or parent category, to ensure the values shown at your chosen default level never

roll up to a higher level.

Note: If you specify a default category that will not be visible to some users because of a user class

security setting, the default category actually shown is the highest available category in the hierarchy

for that user. The default view for that user will show NA for all measures if the new default category

has the Hide the Value option enabled.

Steps

1. Open the diagram view of your model, select the category whose values you want to show

when the cube opens, and from the right-click menu, click Set as Default Category.

2. Still in diagram view, right-click either the root or the parent category for your scenario categor-

ies, click Properties, and on the General tab, confirm that the Hide the Value check box is

selected.

3. Repeat for each scenario dimension.

Note: There is no limit on the number of scenario dimensions in a cube. As long as it is neither

a time dimension nor a measure hierarchy, any cube dimension can be handled in this way.

4. Build the cube and open it in your reporting application.

5. Confirm that expected category values appear, rather than values for the non-selectable category,

for each dimension identified by the scenario dimension icon.

Set Up the Time Dimension: OverviewThe time dimension in your model contains time categories that are meaningful to your users, such

as financial accounting periods or the dates of sales transactions. Transformer supports

● conventional date periods, such as years, quarters, months, weeks, and days

● industry-specific periods, such as 13-week manufacturing periods

● custom periods, such as fiscal years, hours, or minutes

● lunar time periods, such as lunar years or months

● relative time periods, such as year-to-date or previous quarter



Certain time properties are always true: for example, there are always twelve months in a calendar

year, and four weeks in a lunar month. Because these standard properties have been programmed

into Transformer, you can work with time in ways not possible with other dimensions. For example,

you can set up relative time categories to track period-by-period changes in the measures in your

model.

On the dimension line, a time dimension is preceded by this icon .

Working with Time Dimensions

Time dimensions contain date levels arranged in descending order that are usually some combination

of Year, Quarter, Month, Week, and Day. Transformer generates categories for the levels in a time

dimension by applying date functions to the source column that you associate with the time

dimension. The Date Column and the Dimension Name are both required properties for a time

dimension.

You can insert manual levels into a time dimension, but you must specify a date function for each

level so Transformer knows how to relate the categories to their parent levels during the category

generation stage. For example, suppose you insert the manual level Half Year between the source

levels Quarter and Year. After adding the two required categories to your new level, you must also

remember to connect Q1 and Q2 to the first half, and Q3 and Q4 to the second half. Otherwise

your time categories will not generate properly.

Custom Time Dimensions

If the standard date functions don't meet your needs, you can manually create a custom time

dimension. For example, you can add irregular work shifts, or handle time-related data that comes

from more than one source column. You can mix levels that use date functions with levels that

derive their categories entirely from other source columns, such as Sales Promotions for a time

period. However, note that these non-date levels will not be generated if you use the Generate Date

Categories command (Run menu).

Avoiding Multiple Time Dimensions in a Model

For the sake of clarity, you should create only one time dimension in each cube. If you need to

compare values from two date columns, such as the elapsed time between the Order Date and

Shipping Date, use calculated columns, rather than two time dimensions. See "Compare Monthly

Performance in Different Time Periods" (p. 94).

Transformer prevents you from creating more than one time dimension in a single cube that has

time state rollup applied. If you need to track both calendar and fiscal year results, set up alternate

drill-down paths in a single time dimension, converging at a common level such as Month.

Process

To create a time dimension, you can

● use the Date wizard. Transformer prompts you for information about the time dimension and

then creates the dimension for you.



● create a new time dimension and then successively drag the required columns to each level in

it. Remember to set the appropriate date function and other date-related properties for each

level.

● design the dimension before you have any items in the Data Sources list. When you specify

names for the source columns, Transformer warns you that the columns don't exist. When you

click OK and click the Generate Date Categories command (Run menu), Transformer uses

standard date functions to generate categories in the levels of the time dimension without

actually referring to a data source. Later, when the source file is available, you can add it to

the Data Sources list and regenerate the time dimension categories by using data from the source

column. The name you specify for the levels must match the source column name.

● set the Data Class of your time dimension source to Date (Column property sheet), and then

drag it from the Data Sources list to the dimension map. This automatically creates a time

dimension with the standard levels Year, Quarter, and Month.

Define the Date Input FormatSome data sources include information about their columns (sometimes called metadata), while

others don't. Transformer requires information about how dates are formatted in order to correctly

interpret them.

Most date formats are automatically determined during the AutoDesign process. In the time

dimension, if Transformer generates a category named Invalid Dates, it is likely that the date format

is not defined, or not defined properly, for the values in the source file. For example, Transformer

defaults to the YMD format for .csv, .asc, and other flat files. You may have to redefine this default

format for the column in your Transformer model.

Example

As a source file, you are using a comma-separated variable (.csv) file that stores dates in the format

MMDDYYYY. For example, the date value for August 24, 1999 is stored as "08241999". Because

.csv files contain no information about the data that is stored in them, Transformer can only read

these dates correctly if you specify a date format of MDY for the source column containing dates.

Steps

1. Open the property sheet for the column that contains dates, and then click the Time tab.

2. In the Date Input Format box, select the format that matches your source values, and click OK.

Use the Date WizardWhen creating a time dimension containing dates, you can use the Date wizard to automate the

process. The Date wizard issues a series of prompts whose answers provide Transformer with the

information it requires to create a time dimension containing dates.



You can use the Date wizard to construct a time dimension containing either standard calendar or

lunar time periods down to the day level. You can create a time dimension by using the Date wizard

and later make changes to it manually

Steps

1. From the Tools menu, click Date wizard.

2. Type a name for the new time dimension, and click Next.

3. Choose the source column that contains the date values for the new dimension, and click Next.

4. Respond to the remaining prompts, choosing the options that match the type of time dimension

you want to create.

At any time, you can click the Back button to return and change your response to a previous

prompt.

5. When you have entered sufficient information to define the levels in the time dimension, click

Finish.

Create a Time Dimension ManuallyTo create a dimension manually, you drag the columns associated with the levels in that dimension

from the Data Sources list to the dimension map. However, because dates are usually derived from

a single column, Transformer provides the standard date levels for you. You can then add or change

levels as required.

Important Properties for Time Dimensions

When you create a time dimension, several properties are automatically set:

PropertySetting

Time dimension values.The Associations for the time dimension, set in

the Date Level Creation dialog box.

Inclusion of all cube categories, or only those

for which source values exist. This ensures that

relative time calculations are derived correctly.

Whether date categories are always included in

cubes created from the model, set on the Gen-

eral tab of the Level and Category property

sheets.

The time period levels contained in the model,

such as Year, Quarter, and Month.

The date function for each level in the dimen-

sion, set on the Time tab of the Level property

sheet.

How date values appear in the category viewer.The Date Format, set on the Time tab of the

Level property sheet.



PropertySetting

The order in which categories appear in each

date level.

The Order By column for date categories, set

on the Order By tab of the Level property sheet.

Compatible Date Functions in a Drill-Down Path

Ancestor levels in each drill-down path of a time dimension must be compatible. Both Calendar

Year and Calendar Quarter are valid parents of Calendar Months, and both Lunar Year and Lunar

Quarter are valid parents of Lunar Months. Additional valid time periods include 4-4-5 Week

Months, 4-5-4 Week Months, and 5-4-4 Week Months. The convergent level (Weeks or Days) can

have either type of parental level period.

Example

In a sales analysis model, the dates come from the Order Date column in the Orders data source.

When you create the time dimension, Transformer automatically adds levels so that the dimension

looks like this:

By dragging the Order Date column below the levels in this dimension, and setting the appropriate

date functions on the property sheet for the new level, you can manually add a Week level to your

dimension:

Notes

● Use the Date wizard to have Transformer step you through the creation of a time dimension

automatically.

● If you drag a column with a Data Class of Date to the dimension map, Transformer automat-

ically creates the standard time levels in that dimension.

● When Transformer generates a time dimension, it automatically adds several relative time cat-

egories to show period-by-period change in the measures of your cube. You can manually create

additional relative time periods to meet particular reporting needs.

Steps

1. Click anywhere on the dimension line.


3. In the Dimension Name box, type a name for the dimension.



4. In the Dimension Type box, select Time, and click OK.

5. In the Date Level Creation dialog box, select Create Standard Levels.

6. In the Associations box, select the column that contains values for the dates in the dimension,

and click OK.

Transformer creates the new time dimension, adding the levels Year, Quarter, and Month.

7. If you want to add a new level, drag the source column for the time dimension from the Data

Sources list to the new time dimension.

8. Open the property sheet for the new level, and click the Time tab.

9. In the Level Name box, type a name that reflects the date function you plan to apply to this

level (Week, for example).

10. In the Date Function box, select the function Transformer is to apply to this level (Week, for

example).

11. Click OK.

Set Up Fiscal Years, Quarters, and MonthsBy default, Transformer organizes dates in a time dimension according to the standard calendar

year. The Year Begins property controls the date when the year starts. To set up a time dimension

that reflects a non-calendar fiscal year, you change Year Begins from the first day of the calendar

year, January 1, to the first day of your fiscal year, for example April 1.

The label for a lunar fiscal year that spans two calendar years is the initial year. For example, the

default label is 1999 for a lunar fiscal year that starts April 1, 1999 and ends March 31, 2000. You

can change this display to the ending year 2000. See the Enable Lunar Fiscal Labelling option,

General tab, Preferences dialog box.

If you are creating a fiscal year as an alternate drill-down path in an existing time dimension based

on the calendar year, ensure that the two paths converge at a level whose categories coincide exactly.

Steps

1. Open the diagram for the time dimension that is to be based on fiscal years.

2. Open the property sheet for the drill category, and then click the Time tab.

3. In the Year Begins box, type the date on which the fiscal year begins.

Choose a valid date from any year, but ensure that your specified starting date is the first day

in the first week of that year. The default format is YYYY/MM/DD.

4. Click OK.

Set Up Calendar and Fiscal Years Within a Single Time DimensionIn many businesses, measures are tracked over more than one time scale. A common combination

is calendar and fiscal years, where the fiscal year spans different parts of two calendar years.



Typically, calendar and fiscal years converge at the month level. The calendar and fiscal quarters

span different months, but the total number of months in each drill-down path are the same.

When you create alternate drill-down paths, the Order By column for each path must be identical

at the convergence level. Transformer automatically sets the Order By column for the first drill-

down path so that it matches the primary drill-down path. If you add more than one alternate drill-

down path, you must specify the appropriate Order By column for each new path.

Example

In a sales analysis system, the Order Date dimension provides two distinct paths for arriving at the

Month level. One path is based on all years and includes Year, Quarter, and Month. The other

path is based on the fiscal year (which begins in March) and includes Fiscal Year, Fiscal Quarter,

and Month. In Transformer, the dimension map looks like this:

The category viewer looks like this:

Report users can drill down to the month level by either the calendar year or the fiscal year.

Steps

1. If no time dimension currently exists, create one that contains standard calendar time periods,

either manually or by using the Date wizard.

2. In the dimension map, select the level at which you want the calendar and fiscal years to con-

verge.



Note: When connecting alternate drill-down paths in the time dimension, the Year Begins

property (Drill Category property sheet) for an alternate path must be offset by whole units of

the chosen convergence level. See "Alternate Drill-Down Paths in Time Dimensions" (p. 231).

3. From the Edit menu, click Create Drill-Down.

Transformer inserts an alternate drill-down path into the time dimension.

4. In the dimension map, click on the empty area created for the new alternate drill-down path.


6. In the Level Name box, type a name for the parent of the convergence level in the new drill-

down path.

For a fiscal year path connected at the Month level, the parent level name is Fiscal Quarter.

7. From the Associations box, select the column that contains date values for the dimension.

8. Click the Time tab.

9. From the Date Function box, select the function that Transformer is to apply to the level, and

click OK.

10. To add additional levels to the alternate drill-down, select the level you just added and follow

steps 5 through 9 for each new level.

11. Open the diagram for the time dimension.

12. Open the property sheet for the drill category, of the alternate drill-down path, and then click

the Time tab.

13. In the Year Begins box, type the date on which the fiscal year starts, and click OK.

For example, if the fiscal year starts on April 1, type 1999/04/01.

Control How Transformer Splits WeeksIn a time dimension that includes weeks, Transformer makes several strategies available for handling

weeks that span a higher-level time period. You can choose to

● always split the spanning week into two separate weeks, each of which includes the days that

occur in the higher-level time period

Note: In this case, each part of a split week is displayed on the dimension diagram as a separate

week. For example, the week beginning Sunday, June 27, 1999 and ending Saturday, July 3,

1999 is displayed as two weeks: 1999/06/27 and 1999/07/01.

● place the spanning week into the first period, the second period, or the period with the largest

number of days

● split the spanning week into two distinct weeks, but only when a 1-day week is not created as

a result

In a time dimension with multiple drill-down paths, you must follow specific rules with respect to

splitting weeks. See "Alternate Drill-Down Paths in Time Dimensions" (p. 231).



Example

Where the time dimension includes weeks, you must specify how Transformer handles a week that

begins in one month and ends in the following month.

The week beginning on Sunday, March 28, 1999 spans the months March and April. By default,

Transformer splits this week into two partial weeks. It shows the first partial week as 1999/03/28

and the second partial week as 1999/04/01. The first of these partial weeks is a child of the month

March 1999 and has as children the 4 days (starting on March 28, 1999) that occur in March 1999.

The second partial week, named 1999/04/01, is a child of the month April 1999 and has as children

the 3 days (starting on April 1, 1999) that occur in April.

Steps

1. Open the dimension diagram for the time dimension.

2. Open the Drill Category property sheet.

If the dimension contains multiple drill-down paths, open the Drill Category property sheet for

the drill-down path you want to change.


4. From the Partial Weeks box, select the week-spanning rule that Transformer is to apply, and

click OK.

Set Up Lunar Time PeriodsMany systems track time according to the lunar calendar. A lunar year contains 52 weeks of seven

days each. Lunar time dimensions may contain lunar years, lunar quarters, lunar months, and other

reporting periods such as 4-4-5 week months, 4-5-4 week months, and 5-4-4 week months. You

can also add both week and day levels to a time dimension based on lunar time periods.

Because lunar years comprise 52 weeks (not 365 days), you must ensure that the Year Begins and

Week Begins On properties coincide. For example, to set up a lunar year for 1999 that begins on

Friday, January 1, 1999, you must ensure that the Week Begins On property is set to Friday.

Not all combinations of lunar time periods are valid in a dimension. The table that follows illustrates

valid ancestor levels for all lunar time periods:

Valid ancestorsPeriods

NoneLunar Year

Lunar YearLunar Quarter

Lunar YearLunar Month

Lunar Year, Lunar Quarter4-4-5 Week Month, 4-5-4 Week Month, 5-4-4 Week Month

Any higher-level time periodWeek, Day



Notes

● To display the ending year (rather than the starting year) as the label for a lunar fiscal year that

spans two calendar years, select the Enable lunar fiscal labelling option on the General tab of

the Preferences dialog box.

● On the dimension diagram, the default format and source of the lunar date labels for month,

week, and day depend on which label is displayed.

For the lunar month label, the default is YYYY/MM. The month numbering starts with 01,

which corresponds to the first month of your lunar year. The first month is determined by the

Year Begins property, which in turn corresponds to the calendar year. For example, if Year

Begins is set to 19990401, the 04 corresponds to April in the standard calendar, but represents

the first month of your lunar year. On the dimension diagram, the lunar month label for this

date would be 1999/01.

For the lunar week label, the default is YYYY/MM/DD, where DD is the first day of the week.

For the lunar day label, the default is YY/MM/DD. The numbering on the week and day labels

is based on the standard calendar, not on the lunar calendar. Using the Year Begins example

of 19990401, the lunar week labels for the first and second weeks would be 19990401 and

19990408. The lunar day labels for the first and second days would be 990401 and 990402.

● You can use the Date wizard to create lunar time dimensions. Simply choose Lunar Years of

52 Weeks when Transformer prompts for the type of year to include in the time dimension.

Steps

1. On the dimension map, create a new time dimension. Transformer automatically adds the levels

Year, Quarter, and Month.

2. Open the property sheet for one of the levels in the new time dimension, and click the Timetab.

3. In the Date Function box, select the required lunar date function.

For example, select Lunar Year to create a level based on lunar years.

4. Repeat steps 2 and 3 for each level.

Control When Transformer Adds Extra Weeks for Lunar Time PeriodsWhen you use lunar time periods, a year contains 52 weeks of seven days each. A lunar year,

therefore, contains 364 days, which is either one or two days less than a calendar or leap year,

respectively. You can tell Transformer when to use the extra day or days to create an extra week

in the lunar year. Transformer adds this extra week to the last month or quarter of the year. That

way, each lunar year remains aligned with the specified Start-of-year day, while not falling too far

out of alignment with the calendar year.

You can force Transformer to add the extra week when the lunar year becomes 4, 5, 6, or 7 days

out-of-step with the calendar year. Alternatively, if you specify None, Transformer never adds these

extra weeks.



When you create alternate drill-down paths in a lunar time dimension, both the Week Begins On

and the Add an Extra Week settings must match in all drill-down paths if the convergence level is

Week, Lunar Month, or Lunar Quarter.

If the last lunar month in the year has five weeks in a 4-4-5 pattern, the surplus days create an extra

week. The extra week in the quarter is added to the previous lunar month to make a 4-5-5 pattern,

rather than 4-4-6, which is not a valid pattern in Transformer.

Steps

1. Open the dimension diagram for the lunar time dimension you want to modify.

2. Open the property sheet for the drill category that you want to change, and then click the Timetab.

3. From the Add an Extra Week box, select the setting that you want applied with the extra days

of each year.

Limit the Range of Dates Included in the ModelWhen you create the time dimension, you can limit the range of acceptable dates so that categories

that are irrelevant to your users do not appear in PowerPlay.

When Transformer encounters date values outside your specified range, it generates an Early Dates

or a Late Dates category, or both, depending on when the out-of-range dates occur. In addition, if

Transformer encounters dates that are neither early nor late, but can't be placed within the specified

range, it generates an Invalid Dates category.

You can use the Generate Date Categories command (Run menu) to generate date categories and

further refine the range of generated dates.

Example

A source file contains data for the years spanning January 1, 1996 to December 31, 1999. You set

the absolute range of date values to match that range. Transformer encounters a few dates in the

source that fall outside that range (due to invalid data entry, for example) and generates an Early

Dates or Late Dates category in the time dimension.

Steps

1. Open the property sheet for the time dimension whose absolute range you want to change, and

then click the Time tab.

2. In the Earliest Date and Latest Date boxes, type the dates that represent the lower and upper

boundaries of the range respectively, and click OK.



Generate Date CategoriesAlthough Transformer updates the categories in the time dimension whenever you generate categories

or create a cube, you can order Transformer to generate categories for the time dimension alone.

Steps

1. On the dimension line, click the time dimension for which you want to generate categories.

2. From the Run menu, click Generate Date Categories.

3. In the Time Range box, type the oldest and most recent dates to include in the time dimension.

By default, Transformer uses the Earliest Date and Latest Date settings you have specified for

the dimension.

Format Date ValuesBy default, Transformer shows dates on dimension diagrams in a YYMMDD or YYYYMMDD

format. You can change the way dates appear in Transformer diagrams and in PowerPlay reports.

Steps

1. Open the Level property sheet for the date level whose format you want to change.

2. On the Time tab, click Modify Format.

3. If the format you want is listed in the Format Codes box, select it, and click OK.

4. To create a new format, edit the entry in the Code box (p. 204), and click OK.

Set Up a Custom Time DimensionBy using the predefined date levels and functions available for calendar and lunar time, Transformer

automatically determines the relationships between various date categories. However, you can

customize your time dimension to track measures over time periods that Transformer does not

directly support, such as

● work shifts and hours within shifts

● project phases and timed sub-phases

After you have created a custom time dimension, you can set up relative time categories for the

periods in the dimension. For example, if you are tracking the number of patients monitored during

nursing shifts, you could find the number of patients monitored in the same hour, previous shift.

Example

In a system of electrical generating stations, temperatures for various pieces of equipment are

monitored every two hours, and the data is captured in a database. A cube containing this inform-

ation is incrementally updated every eight hours so that the most recent temperature readings for

critical equipment are available to the maintenance crew on duty. The cube tracks temperatures

only at the Shift level and the Hour level. Each work day has three 8-hour shifts.

The source data for the equipment readings contains the following columns.



TEMPEQUIPMENTLOCATIONPLANTHOUR/MINSHIFT

235Primary

Boiler

6ELPlant 01020001

237Primary

Boiler

6ELPlant 01040001

233Primary

Boiler

6ELPlant 01060001

235Primary

Boiler

6ELPlant 01080001

228Primary

Boiler

6ELPlant 01100002

232Primary

Boiler

6ELPlant 01120002

231Primary

Boiler

6ELPlant 01140002

233Primary

Boiler

6ELPlant 01160002

...

Steps

1. Click anywhere on the dimension line.

2. From the Edit menu, click Insert Dimension to add a new dimension.

3. In the Dimension Name box, type a meaningful name for the new dimension.

4. In the Dimension Type box, select Time, and click OK.

5. In the Date Level Creation dialog box, select Do Not Create Levels, and click OK.

6. In the Data Sources list, select the column that provides date values for the dimension, and drag

the column below the dimension you just created.

7. Open the property sheet for the time level you just created, and assign properties as required.

For example, you can change the name of the level and, on the Time tab, change the Time

Level Ranking.

8. Click OK.

9. Repeat steps 6 through 8 for each time level you want to add.



Set Up Relative Time Categories: OverviewSome of the most commonly requested reports in any organization are period-over-period perform-

ance reports and trend analyses: how sales in the current period compare to sales from previous

periods, or how last year's budget compares to projections for next year. By including the most

commonly required relative time categories in your cubes, you avoid the need to calculate them

afresh, in every PowerPlay report.

Relative Time Category Types

Your model can include up to three different types of relative time categories:

● single-category periods, such as the same month, prior quarter or next year

● to-date periods, such as life to date, year to date, or quarter to date

● N-period running totals, such as a 2-week total in the previous month

You can also create custom to-date and N-period relative time categories, to span specified time

ranges.

Default Relative Time Categories

The following relative time categories are automatically inserted into your model:

● Current <period>

● Last <period>

● <Higher-level periods> To-Date Grouped, including

<Higher-level periods> To-Date

<Prior higher-level periods> To-Date

<Higher-level periods> To-Date change (calculated categories)

<Higher-level periods> To-Date growth (calculated categories)

The periods depend on the date levels defined in your model. The grouped special categories only

appear in PowerPlay when the time dimension contains one or more levels higher than the current

period. In Transformer, the relative time periods appear in the category viewer as special categories,

highlighted in green, below the other drill-down paths in the time dimension.



Notes

● You can delete the default relative time categories and add your own custom relative time cat-

egories.

● When using relative time categories, ensure that you set the Inclusion property to Always Include

for each level in the time dimension, so that Transformer can calculate the correct relative date

values.

● When creating a custom relative time category, start with the closest built-in choice and change

it to a Custom relative time category. Then you need only change a few settings to customize

the category.

Set the Current Period AutomaticallyTo ensure that Transformer uses the correct date column to set the current period you can choose

which column in the Data Sources list should be used to set the current period. If two or more data

sources contain columns that provide date values, ensure that only one of them sets the current

period by clearing the Sets the Current Period box in the other property sheets.

If none of the data sources is set to use its date values to set the current period, Transformer sets

the current period to the latest date encountered among all data sources.

The current period is set immediately after categories are generated (and before the creation of any

cubes), provided that the Automatically Set Current Time Period box on the Dimension property

sheet is selected. Before creating cubes, Transformer updates any relative time categories in the time

dimension.

For models based on multiple data sources where more than one source contains date values, you

can have Transformer automatically set the current period to the latest date by clearing the Sets the

Current Period box on the property sheet for every data source.

Steps

1. In the Data Sources list, successively open the property sheet for each data source containing

a date column for the time dimension.

2. On the General tab, clear the Sets the Current Period box for all but the one you want Trans-

former to use to set the current period.

3. Open the property sheet for the time dimension.

4. On the Time tab, select the Automatically Set Current Time Period box, and then click OK.

Set the Current Period ManuallyThe current period for a time dimension, if set, is shown on the dimension property sheet. The

Automatically Set Current Time Period box on a time dimension property sheet determines whether

Transformer automatically sets the current period for the dimension or whether you must manually

set the current period. When Automatically Set Current Time Period is cleared, you can set the

current period to any category in the time dimension.



When setting the current period manually, you generally select a date category that appears at the

lowest level in the time dimension, for example, under a Week level. You can, however, select a

category that is not at the lowest level. When you do, Transformer remembers your selection, and

after generating categories (and before the creation of any cubes), it traverses the dimension until

it finds the latest date category at the lowest level below the selected category. This feature can be

useful, for example, when you want the current period set to the latest date in a year, rather than

to the latest date in your data.

If the current period changes, Transformer immediately updates all relative time categories in the

dimension.

Steps

1. Open the property sheet for the time dimension.

2. Clear the Automatically Set Current Time Period box, and then click OK.

3. Open a diagram for the time dimension.

4. Select the date category that you want to be the current period.

5. From the Diagram menu, click Set Current Period.

Transformer sets the current period for the dimension. The current period is updated on the

time dimension property sheet.

Measure Past Performance for a Single Time PeriodOne of the most frequently requested reports is one that shows change from one time period to

another. This report lets users gauge current performance in light of previous performance and

isolate growth patterns.

Example

In a sales analysis system, the current period is automatically based on the most recent data available

each time new data is added to the model. The system monitors sales revenues and costs for product

lines. Trends from month to month, quarter to quarter, and year to year are of particular importance.

By using the relative time feature, you can set up categories that represent last month, last quarter,

and last year, as well as the same month of the prior quarter, same quarter of the prior year, and

same month of the prior year.

Steps


2. Position the pointer over the right side of the root category for the dimension.


3. Drag the pointer to the right of the root category.

Transformer creates a new relative category and opens its property sheet.

4. In the Category Code box, type a name for the new category.



5. Click the Relative Time tab.

6. In the Relative Time box, select the relative time period that represents the single period for

which you want to create a relative time category, and then click OK.

For a Year-Quarter-Month time dimension, the available list includes the following:

● Current Month

● Last Month

● Last Quarter

● Last Year

● Same Month, Prior Quarter

● Same Month, Prior Year

Note: Select Custom if you want to override any of the default detail selections on the Relative

Time tab. For example, you can specify the basic approach, a different target time period for

your new category, a new context in which the period will be reported, and the corresponding

offsets, to track values that apply to a period other than the current one.

Measure Past Performance To DateOne of the most frequently requested PowerPlay reports is that of measure growth in the context

of a specific time period. This report lets users compare current performance to past performance.

If you create a Custom relative time period, you can override the default detail selections (Relative

Time tab). Specify the basic approach, the target time period (the period for which you want to

create a new category), the context in which the period is to be reported, and the corresponding

offsets (if for a period other than the current one).

Example

A model is used to monitor sales revenues and costs for product lines. The current period is auto-

matically based on the most recent data available each time new data is added to the model. Trends

in the way these measures accumulate within months, quarters, and years are of particular

importance. You set up categories that represent month to-date, quarter to-date, year to-date, and

life to-date. You also set up relative time periods that measure to-date periods in either the past or

the future, such as Quarter To-Date, Prior Quarter.

Steps




3. Drag the pointer to the right and release it to the right of the root category.


4. In the Category Code box, type a name for the new category.




6. In the Relative Time box, select the relative time period that represents the to-date period for

which you want to create a relative time category, and then click OK.

For a Year-Quarter-Month time dimension, the available list includes the following:

● Quarter To-Date

● Year To-Date

● Life To-Date

● Quarter To-Date, Grouped

● Year To-Date, Grouped

● Quarter To-Date, Prior Quarter

● Year To-Date, Prior Year

Note: Select Custom if you want to override any of the default detail selections on the Relative

Time tab. For example, you can specify the basic approach, a different target time period for

your new category, a new context in which the period will be reported, and the corresponding

offsets, to track values that apply to a period other than the current one.

Measure Past Performance Over Several Time PeriodsIn some cases, you may want to set up relative time categories that span several specific time periods

at some point in either the past or the future. Transformer has several built-in relative time categories

to help you do this. If you have an unusual reporting period that is not covered by the automatically

created relative time categories, you can create a custom category.

If you enter positive numbers for either the Target Offset or the Context Offset, then your model

must include source columns with time periods later than the current period so that Transformer

can use them in positive (future) contexts.

Example

In a sales analysis system, relative time categories are used to track sales for prior months, quarters,

and years. However, some users also want to track sales over six-month periods prior to the current

date in both the current year and the previous year. You set up an N-period running total category

that spans the six months leading up to the current month, this year, and last year.

Steps








4. In the Category Code box, type a meaningful name for the new category, such as Previous 6

Months.


6. In the Relative Time box, select Custom.

7. In the Basic Approach box, select N-Period Running Total or N-Period Running Total

(Grouped).

Grouped lets you easily create a series of categories spanning different ranges of time.

8. In the Number of Periods box, type the number of periods to include.

For example, if you are creating a 6-month running total, type the number 6.

9. In the Target Period box, select the type of period for which the N-period running total will

be kept.

For example, if you are creating a 6-month running total, select Month.

10. In the Target Offset box, type a number that reflects an offset for the Target Period relative to

the current period.

For example, if the current period is December 1999 and you want a 6-month running total

up to (ending) November 1999, type -1.

11. From the Context Period box, select a time period one or more levels higher than the Target

Period, within which you want to calculate the N-period running total.

For example, if you are creating a running total of 6 months, create the total within the context

of Year.

12. In the Context Offset box, or in the Context Range box if you selected N-Period Running Total

(Grouped) in the Basic Approach box, type the number by which the Context Period will be

offset when the N-period running total is created.

For example, if you are creating a 6-month running total relative to a Target Period of last

year, type -1.

Set Up Future Relative Time ProjectionsIf your source data contains forecast values, you can build relative time categories such as Next

Quarter or Next Year into your model, and report future projections for the applicable measures.

Because the forecast values project beyond the current period into the future, you must ensure that

the current period is not set to the latest date in your model.

Example

In a sales analysis model, the current period is set to December 31, 1999. However, a data source

containing sales forecasts for all four quarters of 2000 is also included in the Data Sources list for



the model. By creating relative time categories with positive offsets, you can create Next Year and

Next Quarter projections.

Steps

1. Open the dimension diagram for the time dimension.





4. In the Category Code box, type a meaningful name for the new category, such as Next Month.


6. From the Relative Time box, select Custom.

7. From the Basic Approach box, select the option that matches the kind of projection you want.

Depending on the Basic Approach, different controls appear on the Relative Time tab. In the

example below, the time dimension contains Years, Quarters, and Months, and the current

period is set to December. Below are some possible Relative Time tab entries.

SettingsBasic Approach and Category

Target Period = MonthSingle Category: Next Month

Target Offset = 1

To-Date Period = YearPeriod To-Date Total: Year to Date, Next Year

Target Period = Month

Target Offset = 0

Context Period = Year

Context Offset = 1

Number of Periods = 6N-Period Running Total: First 6 Months, Next

Year Target Period = Month

Target Offset = -6

Context Period = Year

Context Offset = 1

8. After you have set all of the properties for the relative time category, click OK.



Compare Monthly Performance in Different Time PeriodsIf your users want reports that show monthly performance in one time period compared to another,

you can add the extra time dimension in Transformer by using the calculated columns feature. That

way, you need not touch the date columns in your source data.

Example

A user wants to see how monthly performance compares between 1999 and 2000. You create a

cube that yields the following PowerPlay chart:

Steps

1. Add a calculated column to the model.

See "Define a Calculated Column" (p. 61).

2. When prompted to select a data class in the Column Data Class dialog box, select Numeric.

3. In the Calculation dialog box, enter the following formula:

month(<Date>)

where <Date> is the column used to build your time dimension.

4. Drag the calculated column from the Data Sources list to the dimension map to create a new

dimension.

5. Add the required levels, dimensions, and measures to your model and then create the cube.

Set Up Special Categories: OverviewA special category groups regular categories from any level in a dimension, without regard for their

normal hierarchical organization. Special categories are unstructured and, unlike the categories in

an alternate drill-down path, they must be maintained manually. However, PowerPlay can be set

up to show these categories in all drill-down lists, below the regular categories in the same dimension.

Notes

● If a cube is based on an apexed view, whether directly or indirectly by means of a cube group,

the special categories become children of the apexed category.



● By default, PowerPlay summarizes or rolls up measure values in special categories; however,

you can disable the rollup option.

● Depending on your PowerPlay edition, portions of this topic may not apply to you.For more



Create a New Special CategoryTo highlight important data in the model, you can create special categories.

In the dimension diagram, the children of the special category appear both in the main hierarchy

and in the special category branch. Any changes to a regular category within this group are reflected

in both instances.

Example

Your model has a dimension that organizes all of your products by Product Line and Product Type.

You want to group your most promising products into a special category. These include Star Gazer-

2 tents, Moonbeam sleeping bags, SunShelter-15 sunblock, RiverKind bio-friendly soap, and all of

your Water Purifiers. You create a Best Bets category and drag the above categories, in turn, from

the Product and Product Type levels to your new special category.

Steps

1. Open the diagram for the dimension in which you want the special category.

2. Click the right side of the root category, so that the pointer changes to a cross-hair, and drag

to the right.

When you release the button, the Special Category property sheet appears.

3. In the Category Code box, type the name of the special category, and click OK.

4. Connect the special category to one or more regular categories.



Create Special Category LevelsYou can create drill-down levels by using special categories, thereby providing your users with an

alternate drill-down path that uses a structure not supplied by your source data.

Example

A dimension called Region includes your branches from the three countries in the Americas. You

want to create a special category for your most profitable offices in this region, and then subdivide

the category further, by geography.

You begin by creating a special category named Profitable Branches and attach the Americas category

to it. You create the lower level categories East and West and connect your most profitable branches

appropriately: New York and Boston to the East, and Vancouver, Seattle and San Francisco to the

West.

Steps

1. Open the dimension diagram.

2. Click the right side of the root category, so that the pointer changes to a cross-hair, and drag

to the right.

When you release the button, the Special Category property sheet appears.


4. Add another special category to the right of the newly-created special category.


6. Repeat steps 4 and 5 to add more categories.

7. Connect the lowest level special category to the appropriate regular categories.

Disable Rollup on Special CategoriesBy default, PowerPlay summarizes or rolls up measure values in special categories. You can disable

rollup by changing the Category Rollup setting on the Category property sheet.

If the Category Rollup box is cleared for a special category, it signifies that values from this category

are not to be rolled up to the parent category. It is possible for parents to add rollup values from

some children and not from others. If no child categories have Category Rollup selected, the parent



category becomes a placeholder and no values are associated with it in PowerPlay. You cannot

perform rollup on the highest level of a special category.

Steps

1. Open the diagram for the dimension that contains a special category.

2. Open the property sheet for the special category.

3. Click the General tab, and clear the Category Rollup check box.

Generate Categories: OverviewAfter you have added dimensions and levels to your dimension map, you can generate categories

from your source by clicking Generate Categories (Run menu). During category generation,

Transformer reads and analyzes your source data, and builds category trees for the dimensions in

the dimension map. You can then open the diagram for any dimension to add a subdimension or

define a dimension view.

Note: Categories are generated automatically when you create a cube.

Deal with Orphans Of CategoriesTransformer generates Orphans Of parent categories for source data that cannot be placed into a

drill-down path. By default, these categories, called source (or key) orphanages, are suppressed.

Example

A model contains two data sources. One is from an order-tracking system, and supplies data for

Country, Region, Branch, and Customer Number categories. The other contains Customer Number

and Address information. If the first source contains data for last year's orders and the second

contains customer records from the two preceding years, any customer number not associated with

an order in the period covered by the first source generates an Orphans Of category.

Step

● To eliminate source (or key) orphanages, either remove the orphan categories from the source

file or associate them to their proper parents.

Handle Early and Late DatesTransformer automatically generates Early Dates and Late Dates categories if dates from the source

fall outside the range specified for any time dimension. By changing the date range, you can incor-

porate out-of-range dates into your model.

Transformer may also generate an Early Dates or a Late Dates category if the Date Input Format

specified on the Time tab of a property sheet for a date column doesn't match the date format of

the data in the source.



Example

You know that a data source contains regional sales data from 1999 to 2000. You set the Earliest

and Latest dates on the Time tab of the Years dimension property sheet to 19990101 and 20001231

respectively. A week later, the source is updated to include data for 2001. When you generate cat-

egories for your model, Late Dates categories appear. You want to incorporate the new data into

the model.

Steps

1. Open the property sheet for the time dimension, and click the Time tab.

2. Type date values for Earliest Date and Latest Date as they currently exist in your source, and

click OK.

Note: If your source data represents date values by using two-digit years spanning two centuries,

see "Specify Date Categories in Two Centuries" (p. 57).

3. In the Data Sources list, select the query that provides date values for the model.

4. From the Run menu, click Generate Categories from Selected Data Source.

Handle Invalid DatesIn most cases, Transformer can determine and apply the correct date format for input from a source

column. Transformer automatically generates an Invalid Dates category if the correct input format

cannot be determined or if a problem arises with the source data. You may have to change the

format of the date information in your source file, change the Date Input Format property on the

Time tab (Column property sheet), or, if the data is corrupt, rebuild your source file.

Steps

1. In the Data Sources list, open the property sheet for the column on which the time dimension

is based.


3. From the Date Input Format box, select the appropriate date format, and click OK.


Use a Specific Data Source to Generate CategoriesWhen a model is based on more than one data source, you may find it useful to generate categories

from a specific source rather than from them all. If you are working with structural data sources,

you can select a specific source for category generation as a means of testing your model.

The Generate Categories from Selected Data Source command only affects category generation,

not cube creation. Another way of enabling or disabling category generation for a specific cube is

to use the Timing controls.



Example 1

You are working with information from a large client database, and your model references many

separate data sources. You create a subset of the database to build and test your initial Transformer

model. You want to generate categories from the subset until the model is ready to test with the

complete set of sources.

Example 2

You have a model that uses two data sources: Products and Sales Orders. Because new products

are not added frequently, you can generate categories from the Sales Orders source alone until there

are new products to include in the model.

Example 3

Your model is based on several data sources. One source contains a single column that provides

date values. The time dimension contains a Late Dates category because your source data was

recently updated with results from the last quarter. After adjusting the range of dates for the time

dimension, you regenerate the date categories based on the updated source to see if the Late Dates

category has disappeared.

Steps

1. Select the data source you want to use to generate categories.


Set Up a Manual Level for Unknown CategoriesA manually-created orphanage is a category created in a manual level. After you create an

orphanage, any subsequently generated categories that don't have a position defined in the model

become children of that orphanage. By using a manually-created orphanage, you keep new data

separate until you can connect it to the appropriate category.

Example

Your organization is compiling data on customer satisfaction. You want to analyze customer satis-

faction by service, city, and region. The data is being compiled over a two-week period, so the

source is updated frequently. As you build the model, you want to store new data separately from

the part of the model you are working on. You also want to update the data in the completed areas

of your model.

Steps

1. Open a dimension diagram and select the highest (left-most) level.


Transformer creates a new level.

3. Click on the right side of the drill category, so that the pointer changes to a cross-hair, and

drag to a position under the new level.

When you release the mouse button, the Category property sheet appears.



4. Select Category is an Orphanage, and click OK.

Check the Model for Potential ProblemsTransformer always checks the model during imports, before category generation, and before cube

creation, but you can run a manual check at any time. It is good practice to run Check Model at

key stages in the design process, and again at cube creation time. Certain warnings, including some

related to auto-partitioning, do not appear with the pre-cube checks.

To see troubleshooting tips for particular messages generated by Check Model, in the Index tab of

the online Help type: TR<message number>.

Example

You are building a cube and working with several data sources. As you add new sources, you want

to be sure that the new data is correctly related to the dimension map. Running Check Model before

each category generation alerts you to potential problems.

Step

● From the Tools menu, click Check Model.

Prevent New Categories from Being Added to a DimensionWhenever you generate categories for a model, Transformer detects new categories as it reads data

values from the sources in the Data Sources list. When this happens, new categories are typically

added to the model. If this is undesirable in your current situation, you can prevent Transformer

from adding new categories to a dimension.




Example

If you have created a cube group based on a level in a dimension, you might not want Transformer

to add new categories to the level on which the cube group is based. Adding a category to this level

would cause Transformer to automatically add a new cube to the cube group.

Steps

1. Open the property sheet for the applicable dimension.

2. Select Prohibit Automatic Creation of New Categories, and click OK.



Chapter 4: Identify the Measures You Want toAnalyze

Measures are the numbers by which you gauge the performance of your organization. Number of

Units Sold and Profit Margin are examples of such performance indicators.

A model can contain any number of regular or calculated measures, together with logical groupings

that use measure folders. A regular measure is numeric data in a transactional data source. A calcu-

lated measure is new numeric data that is derived from other measures, functions, and constants,

such as Profit, calculated from the measures Revenue and Cost. A measure folder can group existing

measures into a logical collection and, if needed, can be assigned a calculated value itself. For more

information, see "Define a Measure Folder" (p. 102).

In this illustration, the model contains non-measure columns (the Data Sources list), regular measures

(Revenue, Cost and Order Qty), and a calculated measure (Profit).

Notes

To summarize the data for a measure, you can apply rollups. See:

● "Control How Measures Roll Up: Overview" (p. 108)

● "Consolidate Data: Overview " (p. 120).

● "Define a Measure Folder" (p. 102)

Add a Regular MeasureA regular measure is numeric data in a transactional data source, such as the number of units sold,

the cost of goods purchased, or sales revenue.

Example

The following transactional data source has the regular measures Quantity, Revenue, and Cost.

Each record represents a sales transaction.


MeasuresNon-measures

COSTREVENUEQUANTITYPRODUCTSTOREDATE

4000.005000.00500TR139SQSTORE119990114

1538.462000.00200TR139SQSTORE119990201

3200.004000.00400TR139SQSTORE119990210

Step

● To add a regular measure, drag the column from the Data Sources list to the Measures list.

Define a Measure FolderUsing Transformer, you can create groupings of measures by using measure folders. By using a

measure folder your report users can navigate through various measure rollups and drill down to

view the lower level measures in their reports.

Measure folders can be used to represent logical groupings of lower level measures, or they can be

calculated measures without any lower level measures. If a folder has no calculation and no lower

level measures, it is not added to the cube. The calculation of the folder-level measure does not have

to reflect the total value of the measures contained within the folder. The measures can be independ-

ent of each other from a calculation point of view.

Example

Report users viewing Profit Margin as the current measure folder may drill down to see the lower

level measures: Revenue and Product Cost. The sum of these nested measures is reflected in the

measure folder as a calculation. The detail of the calculated measure is hidden from users until they

drill down.

Notes

● If you create a measure folder that is not a calculated measure and does not contain any children,

the Check Model function shows a warning message.

● Inclusion or exclusion of a measure on the PowerCube properties Measures tab does not cascade

to the measures that exist below it. If you exclude a measure folder in the Measures list, the

lower level measures are still visible. Each measure must be excluded individually.

Steps to Change a Calculated Measure to a Measure Folder

1. Right-click the measure, and then click Properties.

2. Click the General tab, and select the Measure Folders check box.


Chapter 4: Identify the Measures You Want to Analyze

You can also specify a new measure folder and drag other measures into it.

Steps to Add a Measure Folder

1. Right-click the measures pane, and then click Insert Measure Folder.

2. In the Measure name box, type a name for the measure folder.

Define a Measure that Counts CategoriesYou can define a measure that counts categories. For example, you can create a measure that shows

how many customers of each type bought a specific product each month, quarter or year. Based

on a unique level (Customer No.), this count includes all non-missing, non-zero values, but does

not double-count: if the same customer buys a product in two months, the quarterly rollup counts

that customer only once.

In the Rollup tab of the Measure property sheet, you specify an Activity Measure to which the

Category Count applies. Note that this Activity Measure cannot be a before or after-rollup calculated

measure, an externally rolled-up measure, another category count measure, or an allocated measure

if the allocation occurs in the same dimension as the category count and above the counted categories

level. An error message appears if you try to specify any one of these restricted types of measure as

the basis for your category count.

You can base category counts on allocated measures if the allocation uses a constant from a

dimension other than the category count dimension. You should only accept the default Activity

Measure (All Measures in the cube) if you are sure that no categories are missing measures, and no

measure values sum to zero after they are rolled up. A category is counted only if it has a non-

missing, non-zero Activity Measure associated with it.

Note: In Transformer Version 7.3 Maintenance Release 1 and subsequent releases, category counts

were added to the list of measures supported for time-based partitioned cubes. However, some

limitations apply. For more information, see "Define a Time-Based Partitioned Cube" (p. 130).

Steps

1. Click the Measures list to make it active.

2. From the Edit menu, click Insert Measure.

3. In the Measure Name box, enter a name for the measure.

4. Select the Type tab and click Category Count.

5. Select a dimension and level. The level must be in the primary drill-down path of the dimension

you have selected, and must be unique (Level property sheet).

6. Select the Rollup tab.

7. Select an Activity Measure for the count.

Note: An error message appears if you select an unsupported measure type.

8. Define any other attributes of the count, such as the format, and click OK.



The count appears in the Measures list of your model. In the resulting cube and reports derived

from it, you see the number of non-missing, non-zero categories for the specified level and

activity measure.

Define a Calculated MeasureA calculated measure is an expression that derives new numeric data from regular measures, other

calculated measures, functions, and constants.

You can describe the calculation in the Description box on the Measure property so that report

users can see this information by using the Explain function.

After you create a calculated measure, you can control whether the measure is calculated before or

after rollup. See "Control When Measures Are Calculated" (p. 105).

Note: Transformer performs all calculations defined within a cube based on rules of precedence.

Where there is an intersection between calculated categories and calculated measures, calculations

defined for measures are performed first. This is important to note for reports where the calculated

measure and category intersect.

Example

You can create a calculated measure for Percent Profit Margin by subtracting Cost from Revenue,

and then dividing the result by Revenue. You select 0% in the Format tab to express your calculated

measure as a percentage (you need not multiply by a factor of 100).

Calculated MeasureRegular MeasuresDimensions

PROFIT

MARGIN

COSTREVENUEPRODUCTSTOREDATE

20%4000.005000.00TR139SQSTORE119990114

25%1500.002000.00TR139SQSTORE119990201

50%2000.004000.00TR139SQSTORE119990210

Steps

1. Click the Measures list to make it active.

2. From the Edit menu, click Insert Measure.

3. In the Measure Name box, type a name for the measure.

4. In the Type tab, click Calculated, and then click the Calculation button.

5. In the left pane of the Expression Editor, expand the Functions, Measures, and Value folders

as needed, select each parameter you want to use, and click the arrow to insert it into the cal-

culation expression (right pane of the editor).



6. When the expression is complete, click OK.

7. Type or select the other features you want for the measure, such as its format, and click OK.

The new measure appears in the Measures list.

Define an If-Then-Else Calculated MeasureIn Transformer Version 7.3 and subsequent releases, your models can include calculated measures

that use if-then-else expressions. For example, you can define the following conditional expression:

IF ("Net Income"<100000) THEN ("Gross Profit"*1.25)ELSE NULL

You can also define an if-then-else calculation to avoid division by zero:

IF ("MEASURE_VALUE_A"=0) THEN ("MEASURE_VALUE_B"=NULL) ELSE ("MEASURE_VALUE_B"/"MEASURE_VALUE_A")

Steps

1. In the Expression Editor, begin your calculated measure definition by clicking the if operator.

An opening parenthesis is inserted next.

2. Enter the rest of the expression as follows:

● Type a conditional expression that resolves to True or False.

● Double-click a closing parenthesis.

● After the automatically inserted expression "then (", type the result to show if the boolean

expression is True.

● Double-click a closing parenthesis.

● After the automatically inserted expression "else (", type the result to show if the boolean

expression is False.

● Double-click a closing parenthesis and click OK.

If your expression is valid, the definition is saved for that measure.

3. If an error appears, try again, using a mathematically correct form.

Tip:Unlike calculated columns, calculated measures support isnull( ) expressions that resolve

to True, which is the condition that arises with a missing value. You can use this capability to

avoid divide-by-zero overflow errors.

4. Build the cube, open it in your reporting application, and confirm that the results correctly

reflect the if-then-else condition you were trying to model.

Control When Measures Are CalculatedWhen you use calculated measures, you can have the measure calculated before or after rollup.

Values for measures calculated before rollup are based on source values instead of summed values,



and are rolled up separately. The Auto-partition feature will not be used if a model contains Before

Rollup measures. (Calculated columns may be used to achieve similar results while allowing auto-

partitioning to proceed.)

Values for measures calculated after rollup are based on the rolled up values.

The following table shows the sequence of operations that take place when calculating measures.

OperationsTiming

1 Regular rollup takes place on all measures.After Rollup

2 Time state rollup takes place.

3 Allocated measures are calculated.

4 Calculations are performed on calculated measures.

1 Calculations are performed as the cube is created.Before Rollup

2 Regular rollup takes place on all measures.

3 Time state rollup takes place.

4 Allocated measures are calculated.

Notes

● If values for calculated measures are generated before rollup, Transformer cannot perform

consolidation. However, you can achieve the same results by using calculated columns instead

of calculated measures. Calculated columns are processed as each row is read into the work

file, and consolidation is possible. See "Define a Calculated Column" (p. 61).

● If values for calculated measures are generated after rollup, Transformer cannot perform time

state rollup. See "Set a Time State Rollup Function for Measures" (p. 111).

● Values for measures calculated after rollup require less storage.

● Transformer cannot perform time state rollup for a measure when there are two time dimensions

in the model.

Steps

1. Open the property sheet for the calculated measure.

2. Click the Rollup tab.

3. In the Regular Timing box, click Before Rollup or After Rollup, and then click OK.



Display Missing ValuesBy default, all missing values display as zeros in PowerPlay. You can have PowerPlay display "na"

instead. That way, missing items aren't misinterpreted as zero values.

Steps

1. Open the Measure property sheet, and click the General tab.

2. In the Missing Value box, click NA.

3. Click OK.

Reverse the Sign in Financial ModelsYou can reverse the sign of measure values for specific categories (but not for root categories) if

the sign would not otherwise make sense to the intended audience. For example, in some accounting

systems, revenue balances are shown as negative values, while expense balances are shown as pos-

itive.

Reversing the sign does not affect the value properties, only how they appear in PowerPlay. However,

your users should be advised that sign reversal can alter the results of calculations done in PowerPlay

Reporter. If you note which categories contain reversed signs in the Description box on the Measure

property, users can see this information in PowerPlay by using Explain.

Example

You are creating reports that use data drawn from an accounting system. Initially, the balances for

assets and expenses are shown as positive, and the balances for liabilities and revenues are shown

as negative. You want to distribute the information to sales managers. To make the information

appear more logical to this audience, you reverse the sign for revenue to positive and the sign for

expenses to negative.

Steps

1. Open the Measure property sheet, and click the General tab.

2. Select the Reverse the Sign check box, and click OK.

This option reverses the values of the measure when the cube is generated.

3. Open the dimension diagram containing the categories you want to change.

4. Open the Category property sheet, and click the General tab.



5. Select the Reverse the Sign check box, and click OK.

6. Repeat steps 4 and 5 for each category you want to change.

Control How Measures Roll Up: OverviewThe Rollup function specifies how Transformer and PowerPlay summarize (or roll up) measure

values from child categories to their parent categories. Summarization occurs when PowerPlay

displays measure values in categories above the lowest level of detail.

PowerPlay uses two kinds of rollups:

● Regular Rollup combines records with identical values in non-measure columns. It is applied

to the time dimension by default, unless Time State Rollup is specified for the measure.

● Time State Rollup combines records with identical values in the time dimension according to

the Time State Rollup function selected. For example, level of inventory has been recorded for

a specific product, at the same warehouse, in the same year, on different dates. If you select

Average, and the Degree of Detail set for the column is Day, PowerPlay sums the identical

records and divides by the number of day categories in the month. However, if you select Last,

PowerPlay selects the inventory value for the last day of the month.

Notes

● A third type of rollup, Duplicates Rollup, is used only when Consolidation is selected. For

details on this type of summarization, see "Consolidate Data: Overview " (p. 120).

● If you select both Regular Rollup and Time State Rollup, Transformer performs the regular

rollup first, and then the time state rollup.

● If the measure is used as a weighted value (as in allocation), only the Sum rollup function can

be used.

● If a model has more than one time dimension, only regular (not time state) rollup can be per-

formed.

● If you describe the type of rollup used in the Description box on the Measure property, users

can see this information in PowerPlay by using Explain.

Set a Regular Rollup Function for MeasuresBy default, PowerPlay automatically summarizes (or rolls up) the values for a measure by totaling

them (Sum function). You can change this Regular Rollup default to Minimum, Maximum, Average,

Count, Count All, Any, or External.

Example

A sales analysis model contains a measure named QTY. In PowerPlay, you want to show the average

monthly quantity sold for each store. You use Regular Rollup (Average function) to obtain the

correct result. Your source data is as follows.



QTYPRODUCTSTOREDATE

500TR139SQSTORE119980101

200TR139SQSTORE119980131

400TR139SQSTORE219980101

600TR139SQSTORE219980131

The following results appear in the PowerPlay/OLAP report.

Notes

● Calculated measures can be computed before rollup (in which case the values are based on

source values) or after rollup.

● For the time dimension, if you select Average for Regular Rollup, PowerPlay sums the records

and divides by the number of records in each month.

Steps

1. Open the Measure property sheet, and click the Rollup tab.

2. In the Regular Rollup box, click a rollup function.

If you select the Average function, you can weight the average by selecting a measure from the

Regular Weight box. However, the weighting measure must have a rollup function of Default

or Sum.

3. Click OK.

Create Cubes with External RollupsIn Transformer, you use the Rollup function to summarize (or roll up) measure values in the cube.

If you have data that you don't want changed in the cube, you can use externally rolled up measures.

By directly assigning measure values to specific data points in the model, you control which measure

values are displayed in the cube.

Externally rolled up measures are contained in externally rolled up transactional data sources. Each

data record uses a category code to reference the category in the model, and each code must be

predictable; that is, it must uniquely identify a specific category in the source file. Otherwise,

Transformer makes the code unique by adding a tilde (~) and a numerical suffix.



If a cube contains more than one instance of the category code Item, it renames the second one

Item~1, the third one Item~2, and so on. These category codes are not predictable. To avoid this

problem, define your category codes in one of the following ways:

● Ensure that all source values are unique in a dimension.

● Use the Calculated Columns feature to create unique categories.

● Edit the Model Definition Language (.mdl) file to make the category codes unique.

Creating a cube with externally rolled up measure values involves

● specifying structural and transactional data sources

● building a model

● creating measures from the measure columns

● mapping the source columns to the dimensions

● defining measures as being externally rolled up

● creating a date, if applicable

● specifying the degree of detail for allocations, if applicable

Notes

● Externally rolled up measures can be used with alternate drill-downs, partitioned cubes, and

special categories. In Series 7 Version 4 and subsequent versions, external rollup is also supported

for time-based partitioned cubes.

● For a special category that has more than one child category, you must provide the externally

rolled up value for each special category; otherwise, the category value will be zero. If the special

category has only one child, you can either supply a value or accept the value taken from the

child.

● In Series 7 Version 4 and subsequent versions, duplicate records are overridden, not summed.

● Category codes that are not in the structural data source are ignored. This outcome is reported

in the log file.

● All records must specify measure values. By default, missing values appear as zeros in reports.

● Consolidation cannot be performed on externally rolled up measures.

Steps

1. Define the dimensions by using structural data sources and then add the transactional data

source to your model.

2. Open the Data Source property sheet for the transactional data source and click the Generaltab.

3. Click the Contains Externally Rolled Up Measure Values box, and click OK.



4. Add the measures to the Measures list.

5. Open the Dimension property sheet for each dimension that is to have externally rolled up

measures applied to it, and click the General tab.

6. In the External Rollup Column box, select the transactional column that contains the category

codes the measure values will map to, and then click OK.

7. For each externally rolled up measure, open the Measures property sheet, and click the Rolluptab.

8. In the Regular Rollup box, click External.

Note: For date dimensions or dimensions without unique source values, specify the column

from which category codes can be assigned in order to make these codes unique in the dimension,

as follows:

● Open the Level property sheet for each level in each dimension with externally rolled up

measures.

● In the Category Code Column box, select the structural source column that will map to

the level.

● If you are using allocations, specify the Degree of Detail on the Column property sheet.

9. Create the cube.

Set a Time State Rollup Function for MeasuresTime state represents the state of a measure at a specific time, for example, the inventory at the end

of the month. The Degree of Detail property is set to the lowest level transaction (or the bottom of

the tree). For example, if your source file contains daily transactions, then your degree of detail

should be Day.

You can use any of the following Time State Rollup functions to combine records containing

identical values in non-measure columns: Minimum, Maximum, First Period, Last Period, Current

Period, and Average. The selected function determines how PowerPlay summarizes the measure

values.

Example

The source data contains warehouse inventory for the middle and end of each month. In PowerPlay

you want to show the inventory at the end of each quarter. You select a Time State Rollup (Last

Period function). Your source data is as follows.

QTYPRODUCTDATE

500TR139SQ19990915

200TR139SQ19990930

300TR139SQ19991015



QTYPRODUCTDATE

300TR139SQ19991031

400TR139SQ19991115

600TR139SQ19991130

250TR139SQ19991215

350TR139SQ19991231

The results after Time State rollup are as follows.

QTYPRODUCTDATE

200TR139SQ199909

350TR139SQ199912

At the year level, PowerPlay displays the last inventory level entered.

Notes

● Regular Rollup is not applied to a time dimension if Time State Rollup is selected.

● You cannot apply summarized views to a time dimension when the model contains a measure

with Time State Rollup applied. See "Consolidate Data: Overview " (p. 120).

● Time State Rollup cannot be applied to a measure if the model has more than one time

dimension.

● If you use Time State Rollup, but more than one record is associated with each data point,

values will be summarized using Regular Rollup. For example, if there are two inventory stock

counts daily, the two records may be summed even though you specified a Time State Rollup

function of Last Period.

● You can use the Calculated Columns feature to consolidate multiple measurements. If so, you

should also set the Regular Timing option to Before Rollup. Otherwise, Transformer performs

the measure calculation after rollup, which yields inaccurate results. See "Control When

Measures Are Calculated" (p. 105).

● If a record is missing, you may get zeros when you apply a rollup. For example, if you select a

time state rollup of Last Period, but your source data does not contain records for every day

of every month, then whenever values are missing for the last day of the month, PowerPlay

displays a zero.

● A time state rollup of Average may also result in zeros, but you can specify that any missing

values display as "na". See "Display Missing Values" (p. 107).



● If you select a time state rollup of Average, the Days in Week affects the results. If the Degree

of Detail is Day, and your Days in Week selection excludes weekends, PowerPlay sums the

records for each month and divides by the number of categories (non-weekend days) in that

month.

Steps


2. In the Time State Rollup box, click a Rollup function.


Time State Weight box. However, the measure associated with the weighted values must have

a Rollup function of Default or Sum.

3. Click OK.

Ignore Null and Missing Values in Specified Time State RollupsIn Transformer Series 7 Version 4 and subsequent releases, you can specify that null and missing

values be ignored when Average and Weighted Average time state measures are rolled up.

However, Transformer only supports this feature if you specify that missing values be treated as

"NA", on the Measure property sheet. Also, you must retain the default setting for First Period,

Last Period, and Current Period. That is, null and missing values cannot be excluded from the rollup

calculations for these measure types.

You can achieve the same result from the command line by enabling the IgnoreMissingValue

property for the appropriate OLE object (objMeasure). When using model definition language

(MDL), set the IgnoreMissingValue keyword to TRUE when you create or update the definition

for a supported measure type.

Steps

1. Open the Measure property sheet and click the General tab.

2. In the Missing Value list box, select NA.

3. Click the Rollup tab.

4. Select the Average and Weighted Average time-state rollups check box and click OK.

Note: If the rollup measure is of type First Period, Last Period, or Current Period, the check

box for the Ignore feature is disabled. Missing (null) data values are always excluded from Min

and Max calculations for rollups, whether they are set by Transformer to display as "0" or

"n/a" (the NA setting).

Set Regular and Time State Rollup TogetherYou use Regular Rollup and Time State Rollup together to summarize measure values in PowerPlay.

When you select both, PowerPlay performs the regular rollup first, and then the time state rollup.



Example

The source data contains the inventory levels for the middle and end of each month in all warehouses.

In PowerPlay, you want to show the total inventory at the end of each quarter. To obtain the correct

result, you define a view that removes the warehouse category, and then use Regular Rollup (Sum)

and Time State Rollup (Last Period). Your source data is as follows.

QTYPRODUCTWAREHOUSEDATE

500TR139SQWAREHOUSE119991115




The results after Regular rollup are as follows.

QTYPRODUCTDATE

900TR139SQ19991115

800TR139SQ19991130

The results after Time State rollup are as follows.

QTYPRODUCTDATE

800TR139SQ199912

Notes

● Transformer cannot perform Time State Rollup for a measure when there is more than one

time dimension in the model.

● You cannot apply summarized views to a time dimension if the model contains a measure with

Time State Rollup applied. See "Set a Time State Rollup Function for Measures" (p. 111).

● For the time dimension, if you select Average for both Regular and Time State Rollup, Trans-

former performs Regular Rollup first. It then sums the remaining consolidated records and

divides by the number of leaf categories in the time period. Lastly, it sums the remaining records

and divides by the number of months in the quarter or days in the month. See "Set a Time State

Rollup Function for Measures" (p. 111).

Steps




2. In the Regular Rollup box, click a rollup function.


Regular State Weight box. However, the weighting measure must have a rollup function of

Default or Sum.

3. In the Time State Rollup box, click a rollup function.


Time StateWeight box. However, the weighting measure must have a rollup function ofDefault

or Sum.

4. Click OK.

Allocate Measures: OverviewThe allocation feature distributes data, specified at a summary level of a dimension, to lower levels.

For example, actual sales revenue may be tracked daily, while sales revenue is forecast quarterly.

Allocation is a useful way of distributing quarterly forecasts to the month and day levels.

The following types of allocation are available:

● From Level (typically by direct association with the data source)

● N/A (do not allocate)

● Constant (allocate a measure as a constant value to all descendants)

● By Measure (allocate to descendants proportionally according to another measure)

You can allocate measures

● over entire dimensions, when the measure appears in a data source that does not reference the

dimension

● over levels in a dimension, when the measure is already specified at a level in that dimension

● over categories in levels, when the measure is specified to the particular level

Process

● Add two or more data sources to a model.

● Define the measures.

● Create a dimension map.

● Allocate measures to dimensions or levels that lack them.

● Check the relationships between dimensions and a measure by using the Show Scope command.

Notes

● You cannot allocate a calculated measure. However, you can allocate a regular measure in

proportion to a calculated measure. See "Control When Measures Are Calculated" (p. 105).



● Models that use multiple data sources (and hence measure allocations) are supported in all

editions of Transformer.

● In some cases, you may want to prevent the automatic rollup of constantly allocated measure

values. In Transformer Version 7.3 Maintenance Release 2 and subsequent releases, you can

set a special variable so that the totals in your summary reports are the same as the constants

specified for each individual member of a subset. Use the steps appropriate to your implement-

ation:

● On Windows, locate the cern.ini file in the installation_location\bin directory, open it in

any text editor, and add the following entry to the [PowerPlay DataServer] section:

DISABLE_CONSTANT_ALLOCATION_ROLLUP=1

● On UNIX, create an environment variable called

PPDS_DISABLE_CONSTANT_ALLOCATION_ROLLUP and set its value to 1.

In Bourne shell, add the line export PPDS_DISABLE_CONSTANT_ALLOCATION_ROLLUP

Note: This change will affect 80/20 suppression, calculated expressions, alternate or special

drill-downs, as well as custom subsets.

Also affected are dimension filtering and alternate and special drilling in Visualizer.

View Measure ScopeMeasure Scope is a color-coded dimension map that highlights relationships between the measure

and levels in a dimension. This map helps you identify where a measure has meaning in models

based on multiple data sources.

The scope of a measure depends on its relationship to the levels in the dimension map:


When the scope level is derived directly, the level takes its category values directly from the measure.

For example:

Date

Year

Quarter

Month

Day

The levels Year, Quarter, and Month are directly associated with the measure, while the lowest

level, Day, is associated with another measure.

Level with Allocated Measures

Scope levels with allocated measures are determined either by a constant or proportionally to another

measure. For example:



All Players

Division

Team

Player

The dimension contains a level named Player, whose measure values are allocated proportionally

to an ancestor level.

Notes

● If a calculated measure is based on two regular measures, the measure scope dimension map

shows the lowest common level. If the lowest common level has been allocated, the level appears

with the same shading (green) for both the allocated measure and the calculated measure.

● You can manipulate the measure scope map in the same way as a regular dimension map. The

manipulations affect all open maps.

● To change the default colors of the measure scope map, click the Dimension Map tab (Preferences

dialog box).

● Use the pop-up menu in the Measures list to see measure scope.

Steps

1. In the Measures list, select the measure whose scope you want to see.

2. From the Edit menu, click Show Scope.

Set Dimension AllocationTransformer automatically allocates a measure to an entire dimension when the source of the

measure does not reference that dimension. By default, Transformer allocates the measure value as

a constant throughout the dimension. You can change the default to N/A (do not allocate) or to

By Measure (proportionally allocate, based on values in another measure).

Models made from multiple data sources, and hence allocation capabilities, are supported in all


Example

Sales revenue is tracked daily by products and regions. Sales forecasts are made annually by product

line. By default, Transformer allocates the forecasts as a constant to the Region dimension.



However, you don't want the forecast sales appearing in that dimension. By using the Dimension

property sheet, you can disable allocation by selecting Do Not Allocate.

Steps

1. Open the property sheet for a dimension that is not referenced by the measure you want to

allocate, and click the Allocation tab.

A list of measures that are candidates for allocation appears.

2. Right-click the Allocation Type for the measure, and select an option.

Note: You can allocate measures to an entire dimension from the dimension diagram by double-

clicking the Root category.

Set Level AllocationWhen a measure applies to only some levels in a dimension, you can allocate the measure from a

level where it applies to all lower levels. You can allocate in a regular dimension or a time dimension.

When you allocate a measure from a level, all categories in the level inherit the allocation type.

However, you can change the allocation type of individual categories. See "Set Category Alloca-

tion" (p. 119).

Models made from multiple data sources, and hence allocation capabilities, are supported in all


Example 1: Level Allocation

Sales revenues are tracked daily by products and regions. Sales forecasts are done annually, by

product line.

You want to allocate your sales forecasts to Product Type and Product Name in proportion to

actual sales revenues. Using the Level property sheet, you set the allocation type for Product Line

to By Measure (Revenue).



Example 2: Level Allocation in a Time Dimension

Starting from the same model as in the example above, you decide to allocate your annual sales

forecasts on a monthly basis, in proportion to actual sales revenues. Using the Level property sheet,

you set the allocation type for Year (the time dimension) to By Measure (Revenue).

Steps

1. Double-click the parent level in the dimension map to open its property sheet.

For example, to allocate measure values to the Product Type and Product Name levels, double-

click the Product Line level.

2. In the Level property sheet, click the Allocation tab.

A list of measures that are candidates for allocation appear.


Note: You can check the relationships between dimensions and a measure by using the Show

Scope command.

Set Category AllocationWhen you change the allocation type for a level, the new allocation type is applied from the categories

in that level to all descendant categories. You can also set the allocation type for an individual cat-

egory.

Example

Some operating costs associated with your product lines are based on sales channel factors, such

as direct sales and catalog orders. Others are based on revenue. You want to allocate costs for each



product line proportionally to each cost factor. Using the Category property sheet, you can set the

allocation type for each product line category.

Steps


2. Open the Category property sheet, and click the Allocation tab.


Suppress AllocationTransformer automatically allocates a measure when

● the measure has no relationship to a dimension; the measure is allocated as a constant value

throughout the dimension

● the measure is allocated from one level to the next lower level; the categories in a level inherit

the allocation type from the higher level

In either circumstance, you can disable automatic allocation so that those measure values do not

appear in your PowerPlay reports. When you disable allocation, Transformer creates data points

with missing values, which are displayed in PowerPlay as zeros, by default. You can have missing

values display as "na" instead. See "Display Missing Values" (p. 107).

Steps

1. Open the dimension, level, or category property sheet.

2. Right-click the Allocation Type, and select Do Not Allocate.

Consolidate Data: OverviewConsolidation uses rollups to combine records with identical non-measure values into a single

record, reducing cube size and shortening access time in PowerPlay. Records may have identical

non-measure values in the following circumstances:

● The source contains transactions with identical non-measure values. For example, two sales of

the same product are made to the same customer on the same day.

● The degree of detail permits it. For example, the Degree of Detail for a column associated with

the time dimension is set to Month, so day values in the source transactions are ignored during

consolidation.

● A dimension is omitted from the cube. For example, two sales of the same product are made

at different stores on the same day. If stores are omitted from the cube, the sales records have

identical non-measure values.

● Categories in the cube are summarized or suppressed. For example, two sales of the same

product are made to the same customer on the same day, but the colors differ. If colors are



omitted from the cube by using either of these dimension view options, the sales records have

identical non-measure values.

In cases 1, 2 and 3, consolidation uses Duplicates Rollup to combine records with identical values

in their non-measure columns. In case 4, unless Time State Rollup is selected (Measure property

sheet), consolidation uses Regular Rollup to combine records with values made identical through

the use of dimension views.

When Consolidation Occurs

In cubes that use auto-partitioning, consolidation occurs automatically.

You can specify the type of consolidation: Yes (with Sort) or Yes (Presort). However, a cube will

not be consolidated if Time State Roll-up is defined for it, or if other roll-up conflicts will result

from the consolidation.

Notes

● To combine records without affecting the cube, you can use Regular Rollups and Time State

Rollups without consolidation. These rollup combinations affect how PowerPlay aggregates

measure values. For inventory records, for example, a Last Period time state rollup tells

PowerPlay to show only the last measure value for each time period. See "Control How

Measures Roll Up: Overview" (p. 108).

● The order in which rollups are done affects cube output: Duplicates Rollup precedes Regular

Rollup.




Example

The source data contains daily transaction records that are used to create the Date and Region

dimensions, as follows.

REVENUEQTYSTOREREGIONDATE

5000.0050Store1East19990105

2000.0020Store2East19990110

4000.0040Store1East19990131

3000.0060Store1East19990201

4000.0030Store1East19990228

3500.0060Store1East19990305

4000.0040Store2East19990315



Because the Degree of Detail is set to Month, Transformer ignores day values. Records for the same

month and store (duplicates, in matching colors) are consolidated (summed).

The results after Duplicates rollup (Sum) are as follows.

REVENUEQTYSTOREREGIONDATE

9000.0090Store1East199901

2000.0020Store2East199901

7000.0090Store1East199902

3500.0060Store1East199903

4000.0040Store2East199903

Your cube has a dimension view where the Store values are summarized to the East level.

After Regular rollup (Average), the monthly records for each Store in the East region are consolidated

again.

REVENUEQTYREGIONDATE

5500.0055East199901

7000.0090East199902

3750.0050East199903

Set a Duplicates Rollup Function for MeasuresDuplicates Rollup specifies how Transformer aggregates records that contain identical non-measure

values, that is, identical category names. The records may or may not have different measure values.

For example, the following records are duplicates:



REVENUEPRICEQTYPRODUCTSTOREDATE

5000.00100.0050Tr139sqStore119990101

2000.00100.0020Tr139sqStore119990101

4000.00100.0040Tr139sqStore119990101

4800.0080.060Tr139sqStore119990101

The default Duplicates Rollup for each measure is None (Regular Rollup); however, you can change

the setting to Sum, Minimum, Maximum, Average, First, or Last.

Example

A source file is very large and contains many records with duplicate non-measure values. You can

choose Average as the Duplicates Rollup function to combine the duplicate records (those with the

same date, store and product, shown in a matching color) and reduce the size of the resulting cube.

The data before rollup is as follows.

REVENUEQTYPRODUCTSTOREDATE

5000.0050Tr139sqStore119990101

2000.0020Tr139sqStore119990101

4000.0040Tr139sqStore119990115

4800.0060Tr139sqStore119990115

The results after Duplicates rollup (Average) are as follows.

REVENUEQTYPRODUCTSTOREDATE

3500.0035Tr139sqStore119990101

4000.0040Tr139sqStore119990115

4800.0060Tr139sqStore119990115

Steps


2. In the Duplicates Rollup box, select a rollup function.

3. If you select Average, you can weight the average by selecting a measure from the Duplicate

Weight box.



4. Open the PowerCube property sheet, and click the General tab.

5. In the Consolidate box, click Yes (with Sort) or Yes (Presort).

Perform ConsolidationTransformer only performs a default consolidation when it will be beneficial. You can request

consolidation by changing the consolidation setting in the PowerCube property sheet.

If values for calculated measures are generated before rollup, Transformer cannot perform consol-

idation. However, you can overcome this problem by using calculated columns, because such values

are always calculated before rollup, and consolidation is again possible. See "Define a Calculated

Column" (p. 61).

Steps

1. Open the PowerCube property sheet.

2. Click the General tab.

3. In the Consolidate box, click Yes (with Sort) or Yes (Presort).

Set Up Drill-through Targets: OverviewYou can specify various drill-through targets in your model, to provide further details or more

timely perspectives on the data. Depending on your edition of Transformer, you can set up drill-

through

● between PowerPlay reports (.ppr or .ppx files) and Impromptu reports (.imr files), to show

users the same or different measures (such as count) to the transaction level of detail, but only

include summary records in the cubes

● between PowerCubes, IBM Cognos 8 or ReportNet reports (.crr files) or Query Definition files

(.iqd and related formats)

● from one cube to another, to overcome row or category limitations or to link cubes that can't

be created from one model because they are stored in different places or updated at different

times

● from a cube to a wide variety of associated files, such as a macros, Word files, or PowerPoint

slides, using OLE automation to launch the required application

● from a PowerPlay cube (.mdc file) to a third-party OLAP source, such as Hyperion Essbase, or

a derivative such as an Essbase linked partition or reporting object. For more information, see

the PowerPlay Connect online help.

Drill Through to Impromptu

You can set up drill-through access to an Impromptu report. When users find interesting trends or

anomalies they can then drill through to the underlying data in the Impromptu report referenced

in the cube.



The advantage of this approach is that the often query-intensive work of ad hoc data analysis is

performed entirely locally, using summarized data in the cube. Only when user analysis reveals

something interesting, such as an exception or a noticeable trend, does the user start Impromptu

to issue a query to the database, which returns a relatively small data set. This approach greatly

reduces the potential for too many queries against the data warehouse or mart.

Drill Through with Impromptu Query Definition Files

If your Transformer model uses an Impromptu query definition file (.iqd) as a data source, Trans-

former automatically adds the corresponding Impromptu report (.imr) to the drill-down list for

each measure. The correct filter is passed automatically, so your Impromptu report displays the

appropriate data based on the dimension line in your PowerPlay report.

For .iqd data sources, the default column name in the model is the database column name. You

may change it.

Drill Through without Impromptu Query Definition Files

If your model isn't based on an .iqd file, or if you want to drill through to destinations other than

.imr reports, you can add the target files to the drill-down list in the Measure property sheet (if you

are restricting the drill-down capability to a single measure) or the PowerCube property sheet (if

the drill-down applies to any point in the cube).

In Transformer Version 7.3 and subsequent releases, you can also restrict drill-through from a

measure or cube by level, to narrow the scope. For more information, see "Add or Remove Drill-

through Targets" (p. 126).

Drill Through to Other Products

You can drill through to destinations other than an Impromptu report. If you want to drill through

from a specific measure, add the destination file to the drill-through list in the Measure property

sheet (Drill-through tab). If you want to drill through from any point in the cube, add the destination

file to the drill-through list in the PowerCube property sheet (Drill-through tab).

If you want to set up drill-through access to remote cubes, you may be able to use the PowerPlay

Enterprise Server administration tool to identify your drill-through targets. However, bear in mind

that this tool identifies each target cube by its label followed by an .mdc extension, not by the full

path and physical file name referenced in Transformer. If you relabel a target cube to make its name

more meaningful to your remote users, remember to update the drill-through access in your model

and recreate the cube.

For more information about setting up drill-through access between cubes and reports created with

third-party OLAP tools, see the PowerPlay Connect online help.

Notes

● If a PowerPlay filter applies to a level that was not declared "unique" in the model, the filter

macro passed to Impromptu contains information for that level and for each ancestor level

until a unique level or the highest level in the dimension is reached. For example, if a category

called Burlington belongs to the level City, which was not declared unique, the filter macro

contains City=Burlington plus the next higher level, State=Massachusetts.



● For time dimensions, measure values are specified only for the most detailed leaf categories.

Categories in the time dimension are expressed as ranges from the first to the last descendant.

For example, the year 2005 would be represented as 20050101-20051231.

● For special and manually added categories, all child categories are handled successively, until

the source level is reached.

Add or Remove Drill-through TargetsIf a model is based on an Impromptu Query Definition (.iqd) file, the corresponding Impromptu

reports (.imr) appear in the drill-through list for each measure in the source. You can remove some

or all of these reports from the Measure property sheets, or add additional reports that are not

associated with the original source. To change targets for all measures in a cube at once, use the

Drill Through tab on the PowerCube property sheet.

Because cubes and other drill-through files may be created in different places at different times,

their data and related information (sometimes called metadata) may not always match the inform-

ation in the original cube. Be sure to check that you are drilling through to the most up-to-date

information.

Examples of Drill-through Combinations

You can create a combination of drill-through targets to accommodate your various needs, as follows:

● Drill through from a higher-level cube, showing the results for an entire product line, to a cube

that has the same measures and dimensions but has additional levels in the Products dimension.

Users can drill through to the detailed cube to analyze results based on different attributes such

as color or model.

● Drill through to a cube having the same degree of detail, but differing dimensions. The drill-

through cube may be separated from the original cube in time and space. For example, your

regional office maintains client addresses in an IBM Cognos Query report. Add this report to

the drill-through list on the PowerCube property sheet so your users can see address details in

PowerPlay when they select any cell.

● Drill through to a cube having the same dimensions, but not all of the measures. For example,

a model based on an .iqd file has the measures Profit Margin, Revenue, and Cost, and the drill-

through report shows all three. You can remove the .imr report from the property sheet of the

Profit Margin measure.

● Drill through to only those reports that are in scope, excluding all others based on a set of

restrictions established in the cube model.

Steps to Add or Remove Drill-through Targets

1. Open the Measure property sheet, and click the Drill Through tab.

Note: Or, if you are setting up drill through for all measures in a cube, open the PowerCube

property sheet and click the Drill Through tab there.

2. If you are setting up drill through for the first time, click the Allow Drill Through for this

Measure box.



3. Click Add, find the target file, and click Open to add it to the custom drill-through list. You

can then select it and click Modify to make changes, or Remove to delete it from list.

4. Optionally, type explanatory text in the Report Description box. Your users can then select

Explain in PowerPlay to see this information.

5. Click OK.

Steps to Restrict Drill-through Targets by Level

1. In the DimensionMap, from the Edit menu, click Show Scope and then select the DrillThroughcheck box.

2. From the first drop-down menu, click the measure or cube on which to restrict drill-through

by level.

3. From the second drop-down menu, click the drill target for which the restriction will apply.

4. In the dimension map, click the level or dimension for which you want to restrict drill-through.

5. Right-click and click Exclude DrillThrough.

The interface highlighting shows which level or dimension is now unavailable, or out-of-scope,

for drill-through. By default, the item appears on a red background: the highlighting used for

a source with missing columns.

6. Build the cube, open it in your reporting application, and confirm that the level or levels you

excluded do not support drill-through.

7. Test that the error-handling is consistent, in your client application. For example, PowerPlay

users who click on the drill-through icon while positioned in an excluded level see the following

message: "There are no drill through targets available. Verify your selection or confirm that

drill through is available at this level.

Note: You can only restrict drill-through targets in PowerCubes. Drill-through operations for

third-party cubes are managed by the PowerPlay Connect component.

Add Drill-through Targets to UNIX CubesIf you have the appropriate edition of Transformer, you can set up drill through to a cube on a

UNIX server. First, you must create a temporary file in the Custom Reports list. Once you have

created the temporary file, you can enter the path of the file you want to drill through to.

Steps

1. Open the PowerCube property sheet and click the Drill Through tab.

2. Click Add, name your temporary file in the File Name box, and click Open to add it to the

Custom Reports list.

3. Select your temporary file from the Custom Reports list and click it once to edit the filename.

4. Type the path to the UNIX server, and press Enter.



Chapter 5: Define and Create PowerCubes

When you select the Create PowerCubes command (Run menu) in Transformer, you create a default

or local PowerCube with no custom properties. You can distribute the resulting binary file, or files

if the MultiFileCubeThreshold setting has been exceeded. These contain all the source data for the

dimensions and levels in your model.

However, to optimize processing in your production environment, we recommend that you define

custom views, and take advantage of efficiencies afforded by cube groups or time-based partitioned

cubes. Experiment until you have a satisfactory balance between cube creation and data retrieval

times.

Try creating

● similar but different models that meet the needs of each target group (perhaps pointing to dif-

ferent source files), then build one focused cube from each of these models

● one model and one cube, but use dimension views to remove unwanted categories and offer

different perspectives on the data

● one large shared cube, but hide certain dimensions from some users by implementing Series 7

user class views

● a group of several smaller, related cubes from one model, but control the measures and details

created in each cube

● a group of smaller member cubes, related across a time dimension, and combined into one

larger time-based partitioned cube

Note: Depending on your PowerPlay edition, some topics in this section may not apply to you. For

more information about the features of Transformer in the various PowerPlay editions, see the

Features and Editions section in the Welcome chapter.

Define a PowerCube ManuallyWhen you want to tailor the properties of a PowerCube to meet the specific needs of users, you

can define the cube manually. Examples of items that you can customize include the location of the

PowerCube, the dimensions and measures to include, and password security.

To create a cube, your model must have at least one dimension with at least one level, and its

Measures list must show at least one measure.

Steps

1. Click in the PowerCubes list to make it active and, from the Edit menu, click Insert PowerCube.

2. In the PowerCube Name box, type a name for the new PowerCube.

3. In the PowerCube file name box, specify a location to save the PowerCube.


4. In the Temporary file name box, type an alternative name for the PowerCube.

5. On each of the other tabs, set the properties you want to apply.

6. Click OK.

Define PowerCube GroupsYou can create a cube group for any level in a dimension, and then add member cubes that include

selected categories or measures. For each cube in a group, you specify the degree of detail to be

included so that your users see only the most relevant data.




Example

Montor Expeditions, a nationwide chain of travel agencies, uses one data source and one model to

generate reports on its sales operations. But, instead of creating one large cube that is accessed by

users in different user classes, a set of smaller cubes are accessed by only one user class each. Senior

managers see total and regional sales figures for each travel product. Regional managers see figures

for their regions only.

Although it takes longer to build several smaller cubes, managers get faster query response times

and each regional cube only contains data relevant to that regional manager.

Steps

1. Click in the PowerCubes list to make it active and, from the Edit menu, click Insert PowerCube.

2. In the PowerCube Name box, type a name for the new PowerCube group.

3. Click the Cube Group tab.

4. In the Dimension box, select the dimension that you want to build the PowerCube group from.

5. In the Level box, select the level whose categories will become the individual cubes in the

PowerCube group.

6. In the Focus of Detail group, select the levels of detail as required.

For more information about each option, see the What's This? Help.

Define a Time-Based Partitioned CubeTime-based partitioned cubes are a collection of child cubes, based on one level in the time dimension,

that together form one large cube. Each member cube is partitioned, or split, at the appropriate

level, such as year, quarter, or month.

Report users can view each cube independently, or access the entire collection of member cubes as

one time-based "virtual" cube. This means that reports can be viewed across the entire time

dimension, or across only one level in the time dimension, such as a specific month.



Time-based partitioned cubes offer a better alternative for updating time-segmented data than tra-

ditional cube groups or incrementally updated cubes. The benefits of using time-based partitioned

cubes include easier management, smaller update windows, enhanced end-user performance, and

the elimination of large cube rebuilds.

Category sorting in time-based partitioned cubes was enhanced in Series 7 Transformer, and support

for external rollups was added for time-based partitioned cubes in Series 7 Version 4 and subsequent

versions.

Example

In this sales analysis example, the time-based partitioned cube group is defined on the Quarters

level and cubes are rebuilt every quarter.

After one year, there are separate cubes for the four quarters of 2005. In Q1 of 2006, you run an

update creating a 2006_Q1 child cube. The control cube has categories in the time dimension for

2005 and 2006, and the text file (.vcd) generates 5 lines, one for every quarter (2005_Q1 to 2006_Q1

inclusive).

Quarters.mdc

2005_Q1.mdc

2005_Q2.mdc

2005_Q3.mdc

2005_Q4.mdc

Quarters.vcd

Implementation Details

Transformer creates and maintains two separate files to manage the time-based cube group:

● a time-based control file (.vcd), an editable ASCII-format text file that references the cubes that

make up the time-based group, and their physical locations

● a control cube (.mdc), which

● contains the high-level metadata about the overall structure of the cubes

● maintains a list of all measures, dimensions and root categories, currency records, and the

entire structure of the time dimension

● contains the control information to gather the data from the member cubes

● serves as the entry point for PowerPlay users to access the time-based partitioned cube

● shares the same filename and location as the .vcd file

Advantages

● Support for rolling time periods. You can manually edit the .vcd file to remove references to

cubes that are no longer required, and drop invalid or out-of-date categories. The control cube

and definition file are automatically updated with the newest categories and cube references.



For more information about implementing rolling time support, see "Customizing a Time-Based

Partitioned Cube" (p. 136)

● Support for slowly changing dimensions. Existing child cubes retain the history, and new cubes

can be easily created by taking advantage of the Category Move capability.

● Faster build times. Time-based partitioned cube build times tend to be much faster when com-

pared with traditional incrementally updated cubes. In time-based partitioned cubes, new data

is typically added in the form of a single partitioned cube, rather than adding data to a large

existing cube. Time-based partitioned cubes also eliminate the periodic need to do a full build,

and this results in shorter down-times for production systems, and a more manageable update

schedule.

● More flexibility. By default, time-based partitioned cubes only relate to one specific level of

time granularity, such as Month. However, you can still reference other time levels in the same

model or cube, if this will improve runtime performance.

● Better performance. When report consumers drill-down into the time dimension, performance

improves because they are accessing fewer cubes. When they reach the level of time granularity

that the cubes are based on, for example January or Q1, performance will improve further

because they are now only accessing a single cube.

Category Inclusion in Time-based Partitioned Cubes

Not all children of a category are found in each member cube. Category inclusion is based on the

time level specified when creating the time-based partitioned cube. If data is outside the specified

time period, it is not included in the time-based partitioned cube.

Ideally, modelers should be able to create time-based partitioned cube groups where every child

cube contains all the relevant or possible categories: the "inclusive" approach. However, this

practice may not always be feasible. For example, suppose new categories are only added to the

latest cube. The sort order is then stored in the individual child cubes, called "disjoint member

cubes", rather than the overall parent cube. Transformer may not have the necessary context to

interpret the new parent-child relationships, or to guarantee proper sorting of categories.

Although Series 7 Transformer and subsequent releases use new algorithms to better handle such

scenarios, unexpected sorting orders may result from any significant change to the model, including

● multiple category deletions

● the addition of many new categories that lack contextual information about their parent-child

and sibling relationships

● use of the Unique Move feature

Sorting Considerations

In your reporting application, a group of time-based partitioned cubes appears to be a single "virtual"

cube, but the data in the individual cubes represents a business that may be changing slowly over

time.

Even if the control cube contains high-level metadata about the overall structure of the cubes, fre-

quent changes to the metadata may affect how time-based partitioned cubes perform. For sorting



to work properly, for example, the changes in the categories must be relatively slow and orderly,

and there must be no significant changes to the metadata categories in the overall model.

The list of children of a category consists of the merged list of all children of that category, from

each of the separate member cubes.

By their nature, time-based partitioned cubes are designed to handle large amounts of time-related

OLAP data in a seamless, efficient manner. As noted above, cube updates and query times are

normally faster. However, the extra query time used to sort across multiple children cubes may

incur a measurable performance overhead. There may also be an increased memory requirement.

The "overhead" varies, depending on such factors as the number of cubes referenced by the query

and the number of child categories per parent being sorted.

Adding New Data

New categories are added to the newer cubes in the normal manner. However, older pre-existing

categories may not have new data associated with them in the later cubes. In such cases, not every

child cube may contain the complete set of metadata (category information, relationships, and

measure values).

If the category lists do not remain stable over time, or if major changes occur between cube updates,

it may be difficult for Transformer to understand and correctly interpret the context of any moved

categories. That is, the integrity of the category hierarchy, or parent-child and sibling relationships,

may not be reflected in each child cube. Use of the Unique Move feature compounds the problem,

because the moved category becomes separated from its context (its parent and siblings). This can

produce unexpected sorting results.

Example 1: Adding New Data

Suppose your model design includes several child cubes and the data added to your time-based

partitioned cube includes new categories that did not previously exist. Because there is an overlap

in the time periods, the overall cube sort order reflects a merging of the categories. More recent

categories are appended to the end of the list, destroying the sort order. The parent context is

unavailable in the child cubes, so the category list cannot include categories missing from the

member cubes.

If your query spans multiple child cubes, the overall sort order is a merging of the categories from

the individual cubes.

Example 2: Adding Categories to Disjoint Member Cubes

Suppose your model design includes disjoint member cubes. The overall category order in the parent

cube is A, B, C, D. Categories A and B appear in one child member cube; C and D occur in the

other.

Previously, when you added a new category, E, to the A+B member cube, the sort order in the

reporting applications may have been A, B, E, C, D. Using the new algorithm, E is correctly recog-

nized as the child of the same parent as A, B, C, and D. Sorting occurs as expected: A, B, C, D, E.

Example 3: Changing Dimensions in Disjoint Member Cubes

Time-based partitioned cubes are optimized to undergo incremental updates in which the metadata

changes gradually over time. They are not designed for cases where significant numbers of child



categories are deleted or moved, using Unique Move for example. Operations such as these remove

historical context, producing disjoint member cubes in which unexpected sort orders may appear.

Slowly Changing DimensionsTime-based partitioning allows you to add data to an existing cube and move existing categories

within a dimension, while maintaining their category code identifications, a capability referred to

as support for slowing changing dimensions. For example, suppose you make the following changes,

over several months:

● Delete a category.

● Add a new category in its place.

● Move the added category to a new parent.

Such changes are sufficiently gradual to preserve the context, or original parent-category information,

from one update to the next. The original sort order is maintained.

Bearing in mind the warnings about unexpected sort order, you can use unique moves in time-based

partitioned cubes if both Unique and Move are selected on the Level property sheet. Any changes

in the model are reflected in the new member cubes.

If you make changes in the model and attempt to build the time-based partitioned cube, a warning

message appears indicating that the model has changed. If you continue with the build, the new

member cube is built reflecting any changes.

Historical information continues to exist in older member cubes; new information is placed in the

new member cubes. It is possible that rollups do not reflect the expected values because they may

be using data from two sources.

Example: Slowly Changing Dimensions

The control cube may be using the older cube data combined with the data stored in the newest

member cube to form the parent value in the time-based partitioned cube. When the model structure

is changed, the older member cubes are locked and are not updated with any new information.

For example, in the crosstab below, Marthe Whiteduck has sales for 2001 and 2002.

When Marthe moves from Vancouver to Ottawa in 2003, the report now displays the historical

data for 2001 and 2002 in Vancouver, while also displaying the new data for 2003 in Ottawa.



Process

When creating a time-based partitioned cube, Transformer creates member cubes as part of a cube

group defined against a level in the time dimension. Transformer then creates a control cube and

the .vcd, combining the member cubes into one large cube. Only one time dimension can be in the

model: more than one disables the Enable Time-based Partitioning check box.

The member cubes of a time-based partitioned cube must each cover a distinct level, such as Year,

Quarter, or Month. They do not have to all conform to the same time period. You can open the

.vcd file and manually edit it to include or exclude cubes. For example, performance can be improved

by adding entries in the .vcd file for cubes that are at a higher level of the time hierarchy, such as

the Quarter or Year level. For more information, see "Customizing a Time-Based Partitioned

Cube" (p. 136).

Restrictions

Time-based partitioned cubes cannot be created if the model already contains a feature that is not

supported with them.

The following considerations pertain to the use of time-based partitioned cubes:

● Only one time dimension is supported. A time-based partitioned cube cannot be defined if more

than one time dimension exists in the model.

● Dimension calculation definitions that create calculated categories are not supported. You

cannot use a calculated category at the time dimension level in a time-based partitioned cube.

● Manually created special categories are supported, but the placement of special categories in a

time-based partitioned cube depends on the type: time-related special categories do not exist

in member cubes, but are available from the control cube; regular special categories are included

in member cubes when they are built initially, but new special categories cannot be added to

member cubes during subsequent updates.

● New regular categories may be added to the member cubes but, if they are, special categories

or links to special categories are not created.

● In Transformer Version 7.3 Maintenance Release 2 and subsequent releases, category counts

in the primary drill-down path are supported, as long as the counted level is unique and an

Activity Measure is explicitly set on the Rollup tab of the Measure property sheet. If this setting



is left at its default (All Measures in the cube), at runtime some measures will be disregarded

and the count will not accurately reflect all categories with non-zero values.

Note that the Activity Measure for the category count rollup cannot be one of the following:

● a before or after-rollup calculated measure

● an externally rolled-up measure

● another category count measure

● an allocated measure, if the allocation occurs in the same dimension as the category count,

above the counted categories level.

● Category inclusion is based on the time level specified when the time-based partitioned cube

was first created. Not all children of a category are included in each cube. Only those categories

that contain data for the given time period appear in the resulting reports.

● In Transformer Series 7 Version 4 and subsequent versions, external measure rollup is supported.

With this rollup enhancement, you can also open some types of time-based partitioned cubes

that you could not open before. Specifically, if the child cube has a user class view based on

the time partition level, and the user class view excludes or cloaks categories of that level, you

can now open the child (member) cube. For example, a time-based partitioned cube is partitioned

on the year level, and it has a user class view in which a particular year and its descendent

categories are excluded. The child cube can be opened, but the excluded categories are not

shown and the value of the time dimension root is returned as missing (N/A or zero).

● Allocations that are performed from a level above the time-based partitioning level are not

supported. However, allocations can be done on categories below the time-based partitioning

level.

● Temporary cube filenames are not supported, as their use could yield incorrect results.

● Drill-through may yield unexpected results. For example, if you attempt to drill through to a

time-based partitioned cube containing data that is categorized differently among the cubes

that form the time-based partitioned cube group, data may appear inconsistent.

Steps

1. On the cube property sheet, click the Cube Group tab and select the Enable Time-based Parti-

tioning check box.

2. Click OK.

3. Select the appropriate time granularity.

Customizing a Time-Based Partitioned CubeThe .vcd file is an ASCII file that resides in the same directory as the control cube and also maintains

the same filename. The .vcd file contains one line for each member cube.



You can customize the time-based partitioned cube to enhance performance by editing the .vcd

with any text editor. For example, you can exclude a member cube from a time-based partitioned

cube by manually editing the .vcd file.

Example

Open the .vcd with any text editor. The following .vcd file contains eight member cubes, each

defined at the quarter level.

cube "2001 Q1" .\PowerCube\2001 Q1.mdc








The path defined in the .vcd file is relative to the location of the control cube, as indicated by the

leading dot in the following example:

cube "2001 Q1".\PowerCube\2001 Q1.mdc

A full path to the member cubes can also be specified as follows:

cube "2001 Q1"C:\Transformer\PowerCube\2001 Q1.mdc

Create a Multi-Level Time-Based Partitioned CubeIt is possible to create a time-based partitioned cube that is based on several different levels of the

time dimension. For example, you might want to create a cube that is based on one year cube, three

quarter cubes and one month cube. You can define three time-based partitioned cubes in the same

model by generating the member cubes for the specific time levels and then editing the .vcd file to

contain the appropriate member cube references.

Example

The following five cubes could now be accessed through one control cube, improving runtime query

performance:

cube "2001" .\PowerCube\2001.mdc




cube "October" .\PowerCube\October.mdc

Add or Remove Member CubesTo add a member cube, you must supply Transformer with the new data records. A new member

cube can be generated and the corresponding control cube and .vcd files are updated.



You can choose to remove a member cube by opening the model in Transformer and excluding the

category so that no member cube will be created or by editing the .vcd file as outlined below.

Steps

1. Open the .vcd file in any text editor.

2. Edit the line referencing the member cube to be deleted.

3. Delete the existing member cube.

Exclude Measures from a PowerCubeYou can customize a PowerCube or PowerCube group by excluding certain measures so they cannot

be viewed in PowerPlay. You can remove them entirely, or you can include them but hide them

from members of specified user classes.

If you choose to hide a measure from certain users, rather than from the entire cube, you can do

so from the User Class tab, in models with authentication enabled. All measure values are retained

in the cube, but the restricted dimensions can't be accessed by the excluded users.

Take care not to exclude measures that are used in the calculation of other measures. Otherwise,

errors may occur when your cube is created or updated.

Example

You are preparing a cube that contains a financial summary of your sporting goods sales. In addition

to 1999 sales, there are measures for product cost and 1998 sales. You want to exclude these last

two measures from the cube.




Steps

1. In the PowerCube property sheet, click the Measures tab.

2. Right-click a measure, and click Exclude.

3. Click OK.

Omit Dimensions from a PowerCubeYou can omit specific dimensions from a cube to customize it for particular users. You can omit a

dimension from the cube entirely, or you can include it but hide it from specifieds.

If you choose to omit a dimension from a user class rather than from the entire cube, you can do

so from the diagram of models with authentication enabled. The dimension is retained in the cube

but cannot be accessed by the restricted users.



Example

A model uses several data sources to provide information about every facet of your sporting goods

business. The source columns are Product, Customer, Branch, and Order, with each having a cor-

responding dimension. You want one of your cubes to focus on product sales performance. Because

you don't need customer data, you omit that dimension from the PowerCube object.

Note: Depending on your PowerPlay edition, topics in this section may not apply to you. For more



Steps

1. Open the PowerCube property sheet.

2. Click the Dimensions tab.

3. Right-click to select the dimension to omit or hide.

4. Click Omit Dimension.

5. Click OK.

Set a Password for a PowerCubeYou can set a password to restrict cube access to authorized users. If a password is defined at the

root node of a cube group, then the same password applies to all cubes in the group. A password

defined for a member of the cube group overrides the password defined at the root group node.

Users can type their password in a Common Logon window to see their reports in any PowerPlay

client application. Administrators either set up Access Manager authentication based on user class

membership or define auto-access to a cube using database signons. Series 7 PowerPlay users should

be reminded to configure their authentication source so that they can access protected cubes at

runtime without being repeatedly asked to identify themselves.




Steps

1. Open the property sheet for the PowerCube or PowerCube group.

2. Click the Output tab.

3. In the Password box, type the password twice, as prompted.

Note: You cannot retrieve a forgotten password.

4. Click OK.



Automatic Checking for Open CubesWhen attempting a cube build, Transformer checks whether the cube is open in another IBM Cognos

application. If it is open, Transformer warns you that the cube is locked and in use, and also prompts

you to select one of the following options:

Close the reporting application using the cube, and

click Yes.

To retry.

Click No.To use a temporary name.

Click Cancel, to end the cube build process.To quit.

Notes

● If Incremental Update was applied to the cube being generated, then the option of using a

temporary cube name is unavailable. The user must close any reporting application presently

accessing the cube or cancel the cube generation.

● For MDL and Series 7 OLE rebuilds, a default temporary file name may be used. If that tem-

porary file name is rejected, Transformer will retry up to 100 times, appending a digit from 1-

100 to the filename each time.

Use Alternate Data Sources for Cube CreationOne way to customize cubes for individual audiences is to change the source data. You can create

two different cubes based on the same model, but alter the content by using different source files.

The structure of the alternate source file must exactly match the structure of the source file on which

the original model is based. The alternate file must have the same columns as the original for it to

be read into the model in Transformer.

Also, the alternate data source must contain measure values. Purely structural alternate data sources

will be ignored during a cube build.

Example

You decide to build cubes for different sales divisions from different source files, but in all cases

the same sales analysis model can be used. You set up several Impromptu Query Definition (.iqd)

files, each one containing filters that result in the retrieval of data for only one sales division.




Steps

1. Use your database access tools to set up alternate source files that are structurally identical to

the source files on which the model is based. Each must contain measure values (transactional

sources).



2. Open the property sheet for the PowerCube that will use the alternate source file, and then

click the General tab.

3. In the Data Source box, select the appropriate column in the alternate source file.

4. In the Source File box, type or browse for the name of the alternate source file, and click OK.

Create Compressed PowerCubesAfter you define PowerCube properties and user classes views, you are ready to create the Power-

Cube. You can create either a specific cube object or the cubes for all cube objects in the PowerCubes

list. Also, depending on the number of categories in your cubes and the bandwidth of your network,

you may want to compress your Series 7 cubes or cube groups before you distribute them.

Because compression occurs after cube creation, you will not reduce the processing requirements,

but distribution will be more efficient. Weigh the following advantages and disadvantages to

determine if this optimization technique is worthwhile.

Advantages of Cube Compression

● Compressed cubes are significantly smaller (typically 20 to 80 percent of their uncompressed

size).

● Compressed cubes automatically return to their uncompressed state when opened in Series 7

PowerPlay.

● You can incrementally update compressed cubes.

● Options such as Partition Status, PowerCube Status, and Check Local PowerCube are still

available.

Disadvantages of Cube Compression

● Compression in Transformer, and decompression in PowerPlay, each take a few additional

seconds. However, if you only compress your cubes to transfer them to a new location, you

can use a macro to decompress them again, as a batch job. For more information, see the

Transformer Macro Reference help.

● Transformer must decompress a cube before updating it, and then compress it again.




Steps

1. In the PowerCubes list, select the cube or cubes you want to create.

2. Open the property sheet for any cube that you want to compress, click the Store this PowerCube

Compressed option (Processing tab), and click OK.

3. From the Run menu, click Create Selected PowerCube or Create PowerCubes.



Create a Test CubeYou can generate categories or create a cube by using a subset of your source data. This can be

useful if your source data is very large and it would take considerable time to create a cube for

testing purposes.

Example

Your model uses three database sources (two structural data sources and one large transactional

data source):

● Customers (9,400 records)

● Products (1,720 records)

● Daily Sales (350,000 transactional records)

You specify that you want a test build created by using the first 350 records from each data source.

Transformer quickly creates a small cube whose categories you can evaluate in PowerPlay.

Steps

1. From the Run menu, click Test Build.

2. In the Number of Records box, type the number of records that Transformer will read from

the source.

3. Click either Generate Categories Only or Create Cube.

Update the PowerCube MetadataWhen Transformer creates a PowerCube, it reads all the source data, generates categories according

to the model structure, and writes the data to an .mdc file for PowerPlay to use.

There are several factors that affect the time it takes to create a PowerCube, including partitioning,

cube optimization, incremental updates, and, above all, the size of the data sources.

There is other data in the model, however, that does not depend on the category generation process.

Sometimes referred to as metadata, this related information includes

● object names

● short names

● descriptions

● drill-through items

● user class views

You can quickly update the related information for an existing PowerCube without affecting the

main cube data.



Example 1

The formula used in a calculated measure has changed to produce a more accurate analysis. You

describe the calculation in the Description tab for the measure so that users can view it in Series 7

PowerPlay by using Explain. You use the Update Selected PowerCube command to update the

measure description in the cube without changing the cube data.

Example 2

Your PowerCube production schedule involves recreating a cube monthly, while your Series 7

Impromptu schedule involves producing a report weekly. Your users need to drill through to one

or several Impromptu reports from certain cells. On the first week of each month, they only need

the first weekly Impromptu report. On the second week, they need the first and second weekly

reports, and so on, until they have a choice of four reports on the last week of each month. Although

you only create the full PowerCube once a month, you can update its related information weekly

by adding the extra Impromptu report for the required measure.

Example 3

The makeup of the user classes in your Series 7 model has changed to reflect a company reorganiz-

ation. Access to your PowerCubes is controlled by Access Manager user classes, so you need to

update the authentication data. You retrieve the latest user class structure by clicking Update Access

Manager Information (Tools menu). You then update the user classes information in the model,

including any user class views, before updating all the PowerCubes to reflect the reorganized user

classes. The remaining data is unchanged.




Steps

1. In the PowerCubes list, select the PowerCube whose metadata you want to update.

2. From the Run menu, click Update Selected PowerCube.

3. In the Update PowerCube dialog box, select the items you want to update.

4. Click OK.

Check PowerCube StatusAfter one or more PowerCubes are created for a model, you can check their status at any time. You

can also apply a filter in the dialog box to focus on the cubes that need attention.

Example

To check the status of a PowerCube, you define a cube group. Use Check PowerCube Status to

confirm that all cubes are successfully created, before you tell the users that the cubes are available.






Step

● From the Tools menu, click PowerCube Status.

Customize Cube Content with Views: OverviewTransformer offers various options to customize your PowerCubes so that your reports include

only the categories of interest to your users.

● Apex omits ancestors and siblings of a category; use it with dimension views and user class

views.

● Exclude omits a category, its descendants, and their data; apply it from the category viewer

and use it with dimension views and user class views.

● Cloak omits a category and its descendants, but retains the rollup values in ancestor categories;

use it with dimension views and user class views.

● Summarize omits descendants, but retains their rollup values; use it with dimension views and

user class views.

● Suppress omits a category from reports based on the cube, but retains its rollup value in ancestor

categories; apply it from the category viewer or with dimension views. For special categories,

Suppress is the only available option.

The Cloak, Summarize and Suppress options do not remove data from the cube; even if many cat-

egories are omitted, there is no corresponding decrease in the cube size.

Process

When you create a dimension view, use one of the above options to customize the view, select it

from the Views list, and then apply it to your PowerCube object.

You create a user class view in much the same way, but you must configure security before you

include them in your model. Only then can you select your user classes from the available list, define

a customized view by using one of the above options (except Suppress), and associate it with a cube

in the PowerCubes list. When a user opens the cube in Series 7 PowerPlay or IBM Cognos 8 Analysis

Studio, the system verifies that the user is cleared to see the secured data, before showing the portion

of the cube defined in the custom view. This method is useful when access to shared cubes is a

security concern.

You can create cubes for different audiences based on dimension views, and then further restrict

access by applying user class views to selected cubes in the set. However, certain combinations may

not work well with your partitioning scheme, thereby lengthening cube build times in Transformer.

One workaround is to create several smaller cubes, each using different dimension views, rather

than a create one large cube that has an unacceptably long build time.






Create a Dimension ViewIf you implement reporting solutions for diverse groups in your organization, you may want to

create individual cubes that meet the needs of each group. To do so, you can create one or more

dimension views, and then apply these to the cube objects in the model.

When Transformer creates the cubes, it uses the applied dimension views to write only summarized

versions of the model categories to each cube. Depending on the number of categories removed by

the summarizing option, the cubes may be noticeably smaller.

Note:You must not base your partitions on the same dimension that you used to create a dimension

view. This strategy can result in conflicts and may halt cube creation. Nor should you use dimension

views as the basis for user class views. This can negatively affect auto-partitioning.

Example

A sales-analysis model is used to create PowerCubes for several managers, each of whom is

responsible for sales in a different region. You create a PowerCube for each manager that contains

transaction details in their region, but summarizes values for other regions. You define a regional

dimension view that summarizes the data for all regions other than the manager's region and repeat

for each manager.

You apply this dimension view to the PowerCube object. The resulting PowerCube contains detailed

data for the relevant manager's region, but only summarized information for the other regions.



Tip: Instead of creating a separate dimension view for each category in a level, you can create a

PowerCube group based on that target level. Choose the method that is easiest to implement and

maintain in your own situation.

Steps

1. Open the diagram for the relevant dimension and, if the dimensions pane is not showing, click

the Dimension tab.

2. Select the dimension, right-click, and then click Add New View.

3. Type a meaningful name for the dimension view, and click OK.

4. Select categories, and successively apply view operations (Exclude, Cloak, Suppress, Apex,

Summarize) in the category viewer until you achieve the degree of summarization you want.

5. Open the PowerCube property sheet, and click the Dimensions tab.

6. Right-click a dimension and select View.

7. Select the newly defined dimension view and click OK to apply it to the cube.

Omit Categories Using SuppressIn situations where a category is not required for reporting purposes, but its ancestor and descendant

categories are, you can suppress the category to hide it. Suppression creates placeholder categories

in hierarchies where no values exist for some categories in a level. The values for the descendants

of a suppressed category are rolled up and retained in the ancestor category. Suppressed categories,

including blank categories, are still included in category counts. Values associated with them are

included in rolled-up measure values.



You can suppress existing (as well as blank) categories in a level by suppressing the level itself.

However, in order to properly support the calculation of relative time categories for the cube, you

cannot suppress or exclude weekend days (Saturdays and Sundays) and their associated measure

values.

In dimensions with alternate drill-down paths, suppressing a category located above the convergence

level in one drill-down path has no effect on categories in the other drill-down paths. You can

suppress categories in different alternate drill-down paths, but you can't suppress more than one

entire alternate drill-down path.

Example

Your company maintains branches throughout the world. Some branches are identified by their

province or state. Others are identified by country only. In the model, the country dimension contains

levels for country, province/state, and branch.

To maintain the structure of the dimension, you decide to suppress categories with no associated

source data. In the resulting report, your users can drill down to subordinate categories that reflect

the current structure of your organization.

Steps


2. In the category viewer, select the categories or the level to be suppressed.

3. From the Diagram menu, click Suppress.

4. Open the PowerCube property sheet and click the Dimensions tab.


6. Select the newly defined dimension view and click OK to apply it to the cube.

Omit Categories Using CloakIn certain situations, it may be desirable to omit one or more categories, along with the descendants

of those categories, while retaining their values for rollup into higher-level categories. Like Suppress,

the Cloak option causes a category to be omitted from reports based on the cube. Unlike Suppress,

all of the descendants of that category are also omitted.



When you apply cloaking to a cube, consolidation occurs by default. If new subordinate categories

are added to a cloaked level, they are automatically included in the cloaked ancestors.

Example

In your organization, the French-speaking manager for Belgium and France also oversees your

Montreal office. You define a special dimension view for this manager that includes the transaction

details for Europe and Montreal (Canada), but that cloaks the values for the two other Canadian

offices, Toronto and Vancouver.

You apply this view to the PowerCube object. The resulting cube contains data relevant to this

manager.

Steps

1. Open the diagram for the relevant dimension and add the dimension and user class view that

you want to cloak.

2. In the Dimension, User Class tab, select your new view.

3. In the category viewer, select the categories or level to be cloaked.

4. From the Diagram menu, click Cloak to omit the unnecessary categories.



7. Select the newly defined view and click OK to apply it to the cube.

Omit Categories Using ExcludeYou can exclude categories from a dimension directly from the Dimension Viewer. You do not

have to create a dimension view.

You need to create a user class view first, to exclude categories based on security.

The Exclude option not only omits categories and their descendants from the dimension, user class

view: it also omits the data associated with those categories. Excluded categories are not included

in category counts, nor are values associated with them included in rolled-up measure values. This

differs from the Cloak, Summarize, and Suppress options, which act on the categories but have no

effect on the data written to the cube.

However, in order to properly support the calculation of relative time categories for the cube, you

cannot suppress or exclude weekend days (Saturdays and Sundays) and their associated measure

values.

In dimensions with alternate drill-down paths, excluding categories from one path excludes the

same categories from all other drill-down paths.



Example 1: Exclude a Category and its Descendants from a Dimension View

You want to exclude Far East, United States, and Mexico categories and their descendants from a

view, and from any PowerCube that uses that view. In the cube created with this dimension view,

report users only see the European and Canadian categories. All of the information associated with

the excluded categories is omitted from the cube.

Example 2: Use Combination Dimension and User Class Views to Hide Data

You can use Exclude in combination with one or more other options, to restrict access by your

regionally-based sales executives, based on their security level, as follows:

● Exclude an unwanted measure, Order Quantity, from all user class views, and from the cube

itself.

● Define separate user class views for each sales executive, based on the Regions dimension, and

exclude the Far East and Americas data from the European Sales Executive's view. Repeat,

restricting access as appropriate for each regionally-based sales executive.

● If you are working with Series 7 user class hierarchies, in addition to Far East and Americas

exclusions inherited from the parent user class, you can create a custom view for the German

team so they can see summarized data for the United Kingdom, cloaked data for other European

branches, and details for German and Swedish sales representatives (the latter being affiliated

with the Frankfurt branch).



If you apply this user class view to the PowerCube object, each sales executive and team member

in a subordinate user class can only see reports containing the regional data defined in their custom-

ized view.

Steps

1. Open the diagram and add the dimension, user class view from which you want to exclude

categories.

2. In the category viewer, select the categories or the level to exclude.

3. From the Diagram menu, click Exclude.




Omit Descendant Categories Using SummarizeIf you create a dimension view and use the Summarize option on a category, you remove the data

associated with the descendants of the selected category, but roll up their measure values to the

summarized level. When you summarize all categories in a level, if new subordinate categories are

added, they are automatically included in the summarized ancestors.

Although applying a summarized view consolidates the records in a cube by default, you should

not use this option as a means of decreasing the size of your cube. Also, you cannot apply summarized

views to a time dimension when the model contains a measure with Time State Rollup applied. See

"Set a Time State Rollup Function for Measures" (p. 111).

Steps

1. Open the diagram for the relevant dimension and add the dimension or user class view that

you want to summarize.


3. In the category viewer, select the categories or level to summarize.

4. From the Diagram menu, click Summarize to omit the descendant categories but retain their

combined rollup value.




Omit Categories Using ApexIf you create a dimension view and use the Apex option on a selected category, the resulting cube

contains only the apexed category and its immediate descendants. The ancestors of the category,

any siblings, and any descendants of these siblings are all omitted. Any special categories that include



descendants of the apexed category are connected to the apexed category as if it were the root cat-

egory.

Note: You cannot apex on a union with an alternate drill path.

Example

Illustrated below is the Regions diagram, before and after the Apex option is applied to the United

States category. PowerCubes with this dimension view contain data for the United States and its

descendants only.

Steps

1. Open the diagram.

2. Right-click the Dimension and select Add New View. Or, right-click the view in the User Class

tab, and select Use Custom View.

3. Enter a name for the new view, and click OK.


5. In the category viewer, select the category you want to apex.

6. From the Diagram menu, select Apex to omit all ancestor and sibling categories.

Tip: To reverse an Apex command, apply it at a higher level.






Set Up a Protected PowerCube: OverviewYou can set up protected PowerCubes to be shared among members of user classes, each with access

to only a portion of the data in the cube. Creating protected cubes is a good strategy when you are

creating cubes on a central network server that is accessible by a large number of report users.

It is important to understand the interaction between Transformer and your Series 7 security com-

ponents, before you start. See "Enable User Security for a Model" (p. 153).

Users who belong to more than one Series 7 user class have the option of opening a cube as an

individual user class or as all user classes.



Features and Editions section in the Introduction chapter.

Example

Your sales analysis PowerCube is accessed by both headquarters personnel and the mobile sales

staff and is updated weekly. To control access so that users see only the most relevant information,

you have created the following user classes in Access Manager, and enabled them in the model.

In Transformer, you set up custom user class views for

● Regional Managers, who see detailed data for their regions and summary data for the other

regions

● Sales teams from Europe and the Americas, who only see the sales data for their own work

regions

Process

❑ In Access Manager, define user classes and users with appropriate signons, and release the

resulting authentication data to all users who require access to the cube.

❑ Have users configure their systems to point to the authentication data.

❑ In Transformer, include user classes in the model and associate user classes with the PowerCube.

❑ Create user class views for the members of each user class.

❑ Create and distribute the protected .mdc PowerCube.



Enable User Security for a ModelIn Transformer Version 7.3 Maintenance Release 2 and subsequent releases, you can view and

select user classes from Series 7 namespaces using the User Class Browser. The User Class Browser

lets you select the user classes you want to include in your model, and lets you work more easily

with namespaces that contain a very large number of user classes by allowing you to work with a

subset of the namespace.

Process

Whichever method you use, the process for creating or updating security information remains the

same.

❑ Define your user classes in Access Manager, and then configure Transformer to point to the

namespace.

❑ In Transformer, specify this configured authentication source in the Model property sheet. (If

you leave the box blank, Transformer reads the authentication data configured as Default).

You also have the option of selecting Include Auto-Accesses in the Model, so users are not

repeatedly prompted for their signon information.

❑ Have Transformer import the security information so that it appears in the User Class tab for

the model.

❑ Repeat the process, to update your model should the security information change.

Note: If user classes are added to a Series 7 authentication source, they appear in the list of

available user classes in the User Class Browser, but you still have to add them to the model

manually.

Steps

1. From the File menu, click New to open the New Model wizard. (For an existing model, you

can open the Model Properties dialog box and click the Authentication tab.)

2. Ensure that the Include security in the model check box is selected and proceed to the namespace

dialog box.

3. Select the Include user classes check box, the Use user class browser check box and, optionally

Include auto-accesses. Ensure that you type the Access Manager configuration name to exactly

match the name set up in the Access Manager Configuration tool. If you leave this box empty,

Transformer uses the Default configuration.

4. In the User Class Browser, expand the left pane hierarchy as required, select the user classes

to import, and then click the Add button or drag the selected user class to the right pane. All

user classes under the selected node will be added to the model.

5. Click OK when you have finished adding security, take note of any warning messages, and use

the wizard to finish setting up your model.

6. Open the Diagram for your model, click the User Class tab, and confirm that your selections

match the security defined in your configured authentication location. They are now available

for use in creating up to 10,000 access-controlled views per model.



Note: There is no automatic synchronization of user classes between Transformer and your

configured source. Therefore, user classes must be manually added or deleted if your authentic-

ation information changes. Use the toolbar or right-click menu options to add, update, or

remove the user classes in your existing model, then reset your views.

Update Authentication InformationWhen user classes are enabled for a model, Transformer reads the configured security data and

shows it in the User Class tab. If the hierarchy of the associated security data has changed, or users

have been deleted, the model gets updated but the associated custom user class views become invalid.

To update the security information in your model, you must reload it from your configured

authentication source, confirm that the required user classes, groups, or roles are selected from the

available list, and then adjust your user class views to reflect the updated data.

To speed up your task, if these are the only model components that have changed, instead of

querying each data source, select the Update Selected PowerCubes command (Run menu), and then

clear all of the check boxes except the one pertaining to user security.

Steps

1. Log on as an authenticated user and open the model for which the authentication information

has changed.

2. From the Tools menu, click Update Access Manager Information, and then confirm that your

User Class views have been updated.

3. Use a similar procedure to add or remove obsolete user classes and their related views.

Set Up User Class ViewsInstead of creating multiple cubes and distributing them to different audiences, you can ensure that

different groups of users only see specified dimensions, measures, or categories by

● defining users, user classes, groups, or roles in your authentication source (configured in Access

Manager)

● setting up custom user classes to control access to data and metadata

● securing the Powercubes or cube groups based on your secured model

User classes are applied hierarchically. When you associate a user class with one or more custom

views in such PowerCubes, descendant user classes are also included in that view.

Example

In Access Manager, you set up user classes based on geography: Europe, Far East and the Americas.

The encrypted security metadata may include user IDs and passwords, user classes, cube connection

information, and any combination of basic, operating system, database, or server signons. For more

information, see the Access Manager help.



In Transformer, you set up the authentication that applies to your model and import the user class

information. You then set up user class views for each user class that needs to have a different view

of the data:

● All users share access to the Finance cube.

● Members of each user class only see financial data for their region. You use the Summarize

command to screen Europe and Far East data from members of the Americas Sales Executives,

Sales Managers and Sales Personnel user classes.

● You then limit the view of Europe and Far East sales personnel to their regional data.

● Within each region, the lowest-level class, Sales Personnel, inherits a subset of the permissions

assigned to the higher-level classes: local Sales Managers, regional Sales Executives and their

Managers.

Steps

1. Open the dimension diagram for the relevant dimension and click the User Class tab.

Note: If user security is not currently enabled for the model, see "Enable User Security for a

Model" (p. 153).

2. Select the security object for which you want to create a view, right-click, and then click Use

Custom View.

3. In the category viewer, select categories and apply view operations (Exclude, Cloak, Apex,

Summarize) until you achieve the effect you want.

Note: If users belong to more than one security grouping, they see all the information that they

are authorized to access (that is, the union of all items, not merely their intersection).

4. Drag the secured object that you want to associate with the cube to the appropriate cube or

cube group in the PowerCubes list.

Block Total Values for Parent Categories with Excluded ChildrenIn Series 7 Transformer, you have the option to block total values for parent categories with excluded

children. In your report, you will see a "denied" value rather than the totals of the non-excluded

children. This prevents you from viewing data that is an inaccurate roll up of only non-excluded

categories.

Missing values take precedence over denied values. Categories will show zero, N/A (not available),

a blank (nothing in the cell), or missing values, depending on how the measures in the cube were

designed to handle missing values. This allows you to distinguish between missing and denied values.

You must exclude categories in the User Class tab of the (Categories) Diagram Viewer before

selecting Block Totals for Parents with Excluded Children check box. You must also ensure that

user classes are enabled.

Example for Series 7

You are responsible for a sales cube with North America divided into Mexico, Canada, and USA.

A user is allowed to see the sales values for Mexico and USA, but not the values for Canada, so



you exclude the Canada category. You can choose to restrict aggregate values, so that the user sees

"Denied" instead of summary values, or continue to show totals for North America. If you continue

to show totals, the user could then deduce the values for Canada, but not the details.

Note: If you use filters, the user sees a value for North America that excludes the contribution for

Canada. In this case users might erroneously conclude that Canada had no sales.

Steps for Series 7

1. Click Show Diagram and select the User Class tab.

2. From the user class list, select the dimension you wish to modify.

3. Select the category or level you wish to exclude.

For more information, see "Omit Categories Using Exclude" (p. 148).

4. In the PowerCubes list, right-click the cube you wish to modify and select Properties.

5. In the Processing tab, select the Block Totals for Parents with Excluded Children check box.

Combine User Class Views with Dimension ViewsYou can add user class views to a cube that already has dimension views. Use this combination

when you must control access to certain information (such as salaries) based on a user's security

classification, but need not restrict access to the remaining data.

We do not recommend using dimension views as the basis for user class views if auto-partitioning

is in place. In very large cubes with complex partitioning schemes, test your design to ensure that

combination views do not negatively impact cube creation time.

Example

In a sales analysis model, a PowerCube group contains cubes for Europe, the Americas, and the

Far East. For each cube, additional protection is imposed by creating user classes that group users

separately from security administrators, for each region. When the cubes and the accompanying

authentication sources are released to users in each region, regional security administrators can

maintain security definitions for their own regions.

Steps

1. Create all the dimension views or PowerCube groups you want in your model.

2. Apply the dimension views to the appropriate PowerCube objects. See "Create a Dimension

View" (p. 145).

3. Create and apply user class views to the objects in the PowerCubes list. See "Enable User

Security for a Model" (p. 153) and "Set Up User Class Views" (p. 154).

4. Create the PowerCubes.



Chapter 6: Perform Model and PowerCube Main-tenance

Your production Transformer models and cubes will probably require ongoing maintenance to

respond to the evolving needs of your report viewers, changes in your data, or a reorganization of

your workplace.




Source Data Updates

If you add new transactional records, your model need not change and your cubes can be increment-

ally updated. If you change the structure of your data by re-ordering or renaming columns, you

may need to redesign a portion of your model. See "Match Model and Source Columns" (p. 159)

and "Move Categories When Source Data Changes" (p. 160).

Model Updates

If you delete a dimension, you won't be able to see any existing reports in PowerPlay. If you delete

some categories, the reports will open but show "error" as the value for each deleted category. If

your source contains records for time categories that don't exist in your model, Transformer creates

the appropriate placeholder category: Early Dates, Late Dates, or Invalid Dates. To prevent

Transformer from adding unwanted categories to a dimension, you can lock it from the Dimension

property sheet.

If you are making changes to your model, you should periodically clean up fragments that can cause

generation errors by saving your model in .mdl (rather than the default .py?) format. If your model

has categories that aren't used, delete them with the Clean House command (see "Delete Inactive

Categories" (p. 161)). If your model contains incrementally updated cubes, you must retain all of

their categories for the cubes to be valid.

Cube Updates

You should alert your report users whenever you change a cube object, because the change may

not be obvious by simply looking at the report.

If you enable the incremental update feature for a cube, Transformer appends the new data to the

existing cube. If you disable incremental update, Transformer overwrites the existing cube. See "Set

Up Placeholder Categories for Cube Groups" (p. 161). To rebuild an incrementally updated cube,

delete the old .mdc file and recreate the entire cube.

Cube Connection Updates

After a PowerCube is rebuilt, you can automatically update the cube connection information,

without affecting your IBM Cognos 8 users, by using the sample script cubeswap.bat or cubeswap.sh,

located in the c8_installation_location/webapps/utilities/cubeSwap directory.


You do not need to use the SDK. For help on the syntax, run the script without any parameters.

For example, before updating the "great_outdoors_company" datasource to point to the file Great_

Outdoors_Company_2.mdc, navigate to the cubeSwap directory and type cubeswap.bat

A help message will explain the syntax, and possible error exit codes:

<URL> <dispatcherName> <dataSource> <newCube>[<userName> <password> <nameSpace>]

Where URL is the IBM Cognos 8 BI Server URL, such as

http://<hostname>:9300/p2pd/servlet/dispatch

dispatcherName is the name of the dispatcher that services the cube to be updated, such as

http://<hostname>:9300

dataSource is the name of the source to be updated with the new cube, such as Go Data Warehouse

newCube is the fully qualified name and path for the new cube, such as

<path>/<cubename.mdc>

Optionally, to use a secured namespace with "Anonymous" disabled, specify a userName, the name

of a valid user authorized to run the cubeswap utility, their password, and nameSpace id.

Possible exit codes are as follows:

0 - Success

1 - Invalid command line arguments

2 - Login Failure

3 - Operation Failed

Assuming your applications are running on default ports, to update the above datasource, type:

cubeswap http://<hostname>:9300/p2pd/servlet/dispatch, http://<hostname>:9300, great_outdoors_company, "D:\Program Files\Cognos\series8\webcontent\samples\datasources\cubes\PowerCubes\En\Great_Outdoors_Company_2.mdc",AAA, AAA, cogs7

Note: You must put the path in quotation marks because there is a space in "Program Files".

A successful connection update results in the following message output:

DataSourceChange properties for dispatcher: 'http://<hostname>:9300' updated successfully

DataConnection String for dataSource: 'great_outdoors_company'updated successfully

Recover a Failed ModelTo help you recover from power failures or similar processing interruptions without having to

recreate the entire cube, Transformer adds checkpoint entries after each major stage in the cube

creation process. These checkpoints are written to a temporary .qy? file in the folder specified by

Model Temporary Files in the Directories tab (Preferences dialog box). Because this file is deleted

when the process ends normally, the existence of a .qy? file indicates that Transformer terminated

unexpectedly.


Chapter 6: Perform Model and PowerCube Maintenance

The next time you start Transformer, you will be asked if you want to see the list of suspended

models. You can either open the model at the last checkpoint before failure and continue to develop

it from that point, or continue from the point where you last saved the file. The .qy? file is then

deleted.

Transformer also writes messages to a log file, which is stored in the same location as your models,

or in a location specified in the Directories tab (Preferences dialog box). It has the same name as

your model, but with an extension of .log. If Transformer is unable to recover automatically from

the processing failure, or if you choose to ignore previous processing and begin again, you can read

the log file to find and correct the cause of the failure.

Notes

● Checkpoints do not work if you use auto-partition as an optimization method.

● If you are using Transformer on UNIX, you can restart a job from the server command line:

● To open a checkpoint file in batch mode, use the -p command line option; for more

information, see "-p .py?_file" (p. 250).

● To ignore a checkpoint file in batch mode, use the -i command line option; for more

information, see "-i .py?_file" (p. 247).

Steps

1. From the File menu, click View Suspended Models.

If a severe error occurred during processing that caused Transformer to shut down, you are

prompted to open the Select Suspended Model dialog box when you next start Transformer.

2. Select the option to open a model, and click OK.

Match Model and Source ColumnsWhen you create a model, the columns in each data source are saved as part of the model definition.

If you re-order, add, delete, or rename the columns in a delimited-text, Impromptu Query Definition

source file or Transformer model, you must update the model by running the Modify Columns

command.

Example

An existing source file has columns named DATE, PRODUCT, SALESPERSON, QTY_SOLD and

PRICE. The sales representatives are reorganized into units, and a UNIT column is added to store

this new transactional data. You run the Modify Columns command to update your model.

Steps

1. Select the source in which the columns do not match the model columns.

2. From the Tools menu, click Modify Columns.

A plus sign (+) appears to the left of all model columns that have been changed.



3. For each changed column, click Match, Add, or Remove, as appropriate.

Move Categories When Source Data ChangesYou can prevent model errors due to changes in your source by designating a unique level. This

tells Transformer that categories in that level are identified by their source values alone, without

reference to their ancestors. However, when you move a category from a unique level, a uniqueness

violation is reported during category generation because the moved category now appears in a dif-

ferent context.

To avoid having to manually restructure the categories in a unique level to conform to the changed

ancestor data, you can specify that such changes be treated as unique moves. Measure values, even

those accumulated under the old structure, are thereafter rolled up the new path to the moved cat-

egories. Unless you make special adjustments in your model (such as implementing calculated

columns to demarcate the move date), historical values move to the new path, and are no longer

associated with the old one.

Note: In certain circumstances, unique move is not supported in time-based partitioned cubes.

Example

You have a model defined, but your source data must change when two staff members, Alessandra

Torta (Milan) and Ellen Shapiro (Amsterdam), switch offices.

If you make the switch in your source data without adjusting your model, Transformer puts the

new categories in their new context (left column, highlighted in red). But the values for each indi-

vidual also remain associated with their former offices, which produces inaccurate results in

PowerPlay.

Unique Move SpecifiedUnique Move Not Specified

FirstnameLastnameCityCountryFirstnameLastnameCityCountry

EllenShapiroMilanoItalyAlessandraTortaMilanoItaly

EllenShapiro

......

AlessandraTortaAmsterdamNether

lands

EllenShapiroAmster

dam

Nether

lands

AlessandraTorta

Suppose you make the switch in your source data and adjust your model to make LASTNAME a

unique level (because the switched personnel can be identified without reference to their ancestor

categories). Two error messages appear when you try to generate categories:

● category Alessandra Torta cannot be created in the Netherlands path because it already exists

in the Italy path



● category Ellen Shapiro cannot be created in the Italy path because it already exists in the

Netherlands path

To avoid these problems, you click both the Unique and Move check boxes in the LASTNAME

Level property sheet, thereby specifying the changes as unique moves. You now see the correct

results (right column, highlighted in red).

Steps

1. Open the property sheet for the level that contains the categories affected by the changed data.

2. In the Source tab, click Unique.

3. Click Yes to the warning message that categories in the level must be recognizable by the source

data alone.

4. Click Move, and then click OK.

5. From the Run menu, click Generate Categories.

6. To verify that the categories moved correctly, select the dimension in the dimension map and

click Show Diagram from the Diagram menu.

Delete Inactive CategoriesIf your organization changes, your model may contain inactive categories that are no longer needed.

A category is considered active if it has been created, updated, moved, or had its properties changed

since a specified date.

You cannot delete inactive categories if incremental updates are defined for the cubes in your model.

If you do so, you will get an error message when you try to create the cube.

Example

Last year, your department was responsible for inspecting 40 homes for the elderly. After a consol-

idation of facilities, you now inspect and report on the operations of only 30 homes. You delete

the categories for the 10 homes that have been closed since the start of your fiscal year.

Steps

1. From the Tools menu, click Clean House.

2. In the Date box, type a date using the format configured for your system or select a date by

using the embedded calendar.

All categories that have been inactive from the specified date to the present are deleted.

Set Up Placeholder Categories for Cube GroupsWhen you define a cube group, you can manually add a placeholder category to the level from

which future cubes will be derived, to allow for additions. Provided the new members have the



same rolled-up data as the existing members, the cube containing the placeholder category will

gather the incrementally updated data that the new member needs.

Example

Your model has a cube group for your sales offices, which you update incrementally. Each member

cube therefore contains historical summary data. Your company is planning to add another sales

office in the future, but details aren't yet available. You want the new office to have access to the

same historical data as the other offices.

Notes

● See "Update Your Cubes Incrementally" (p. 163), "Steps to Perform Auto-Partitioning" (p. 171),

and "Steps to Partition Manually" (p. 173) for important information about incremental

updating and partitions.




Steps to Set Up a Placeholder Category and Cube

1. Open the dimension diagram for the dimension on which the cube group will be based.

2. In the category viewer, insert a category into the level from which the member cube will be

derived.

3. Define the cube group.

In the Processing tab, be sure to select the This Cube is Incrementally Updated check box.

4. Click OK.

5. In the PowerCubes list, expand the cube group to list the members.

6. In the property sheet for the placeholder cube, change the name to New.

Steps to Set Up the New Member

1. Open the dimension diagram for the dimension on which the cube group is based.

2. On the property sheet for the placeholder category, rename the Category Label and Source

Value to match the new data column.

3. On the property sheet for the placeholder cube, rename the cube.

Optimize Performance: OverviewAs you implement larger and more complex PowerCubes, optimizing build times and runtime query

performance becomes increasingly important.

Before you build your model, consider optimizing performance by

● consolidating your source data by summarizing details and removing out-of-date records



● adding multiple CPUs and configuring your data sources to use the multiprocessing option

Because cube size, processing time in Transformer, and access time in your PowerPlay reporting

components are all affected by the optimization method you choose, identify which competing

factors are important to you, and then select the most appropriate methods from the list below:

● Use multiple data sources to reduce the size of each source file and shorten processing time in

Transformer.

● Specify how the cube is processed, either by reducing the number of categories in the cube

(which shortens access time in PowerPlay) or by optimizing querying (which shortens processing

time in Transformer).

● Maximize data consolidation by using the auto-partition feature and by adding a sort step

before records are written to the cube; this decreases cube size and shortens access time in

PowerPlay, but lengthens processing time in Transformer.

● Take full advantage of partitioning to speed up runtime access.

● Use incremental updates so that you only add new data during cube updates, which shortens

processing time in Transformer.

● Subdivide large PowerCubes into multifile cubes to improve runtime query performance.

Update Your Cubes IncrementallyYou may wish to design your model so that frequently updated data is added by means of one or

more separate data sources. You can then perform an incremental update to quickly add this data

to the existing cube.

When a cube is first created, auto-partitioning is used by default, but for incremental updates, auto-

partitioning is not used after the first increment. For subsequent increments, the categories optim-

ization method is used.

The first time you run an incremental update, include all the categories for the levels used in the

partitioning. This ensures the best performance from an incrementally updated cube.

If you add new categories to a partition level in a model, the new categories will be stored in the

parent partition level unless you specify the partition level number.

If you add a large number of new categories, the access time for report users may be adversely

affected because of the increased size of the summary cube. In this case, the model should be

repartitioned.

Over time, if new categories are being introduced, incremental updates reduce the effectiveness of

the partitioning scheme in the cube. This, in turn, degrades performance.

It is good practice to periodically recreate the entire cube, in order to optimize its auto-partitioning

scheme (update weekly, recreate monthly, for example). And whenever you make any structural

changes to your model, you must recreate the cube before performing further incremental updates.



and Editions section in the Introduction chapter.



Limitations of Incremental Updates

Incremental updates can save processing time, but there are a number of factors to consider if you

plan to use them:

● Ensure that the data source file contains only new data; otherwise, data duplicated in the data

source is duplicated in the cubes.

● Some changes that you make to a model can invalidate an incrementally updated cube.

● Don't delete a leaf category that is included in this type of cube.

● Don't use Exclude, Cloak, Summarize, or Apex in a view if they cause leaf categories to

be excluded.

● Don't delete, add, or move a dimension or measure.

● Retain all the data that was used to create an incrementally updated cube so that you can

recreate it if you change the basic structure or content of the cube.

● If you disable the incremental update option, ensure that your model queries the source file

containing all of the data required to recreate your cube from scratch (not a file that contains

only changed data).

● If data that is related to a cube remains unchanged since the last update, you can reduce pro-

cessing time by using the Disable Cube Creation command. This might occur if your model

made use of cube groups, and only a subset of the cubes had new data.

● Use the Update PowerCube command to modify a cube if you have made minor changes to a

model. The existing cube is updated, instead of deleted and replaced, and may take significantly

less time than regular cube creation. You can use this command if the only changes you have

made to a model are the following:

● update object names, short names, labels, or descriptions

● update drill-through reports

● update user classes

Notes

● The cube is not a primary data source. You must retain all of the data used to build an incre-

mentally updated cube so that you can rebuild the entire cube should the structure of your

source data change significantly.

● When you disable the incremental update option, subsequent category generation and cube

creation for that cube completely recreates the cube. If you changed the update timing of the

source files, remember to change it back so that all the required data is queried during cube

creation.

● If you have a large cube that changes its structure frequently, but you do not have time to

recreate the entire cube every processing cycle, alternate between rebuilding and incrementally

updating your cube. When structural changes occur, recreate the entire cube, including historical

summary data. When data changes are only minor, update the cube incrementally.





and Editions section in the Introduction chapter.

Steps

1. Add a new cube definition to the PowerCubes list.

2. Open the property sheet for the new cube, and click the Processing tab.

3. Select the This Cube is Incrementally Updated check box, and then click OK.

4. Create the cube.

Recommendation - Keep Models Populated if Category Codes are Not UniqueWhen the category codes in a model are not unique, Transformer makes them so by adding the

characters ~### to each code, where ### represents the number assigned sequentially to each non-

unique item. We recommend that you save models with categories of this type in their fully populated

form, rather than cleaning out the categories between each cube build.

Also, we recommend that you save the model in both .mdl and .py? formats, so that you have a

backup version if the py? file becomes corrupted. This precaution is particularly important if the

same category repeats itself, or if your model contains blank categories.

Furthermore, if your category codes are not unique, and change after each cube build, reports that

contain explicit member references may no longer run successfully. To avoid this problem, we

recommend that you base your reports on level (children of) references, which will not be affected

by category code changes.

Recommendation - Periodically Clean up Your Model to Optimize Cube ProcessingPeriodically, we recommend that you revisit your design, adjusting your data, data sources, model,

or cube creation choices. Consider making the following improvements to speed up cube creation,

reduce cube size, and improve access times for report users:

❑ Delete or exclude records from the source data if they are no longer needed.

❑ Consolidate transactional data to a higher, summary level using the Duplicates Rollup feature,

and implement optional drill-through targets, if the details are not needed by the majority of

your users.

❑ To improve cube creation time, try changing the order of your structural data sources. Start

with the structural data sources that contain the hierarchical data for the dimensions. Then

add transactional data sources to supply measures for the model, using the minimum number

of columns needed to reference those dimensions.

❑ Ensure that your data does not have any uniqueness violations, if you are using level uniqueness.

Allocate extra time for data source processing to verify that all categories are unique within a

level, or eliminate this step if it is not necessary by clearing the Verify Category Uniqueness

option on the relevant data source property sheet.



You can further optimize cube creation by specifying one of the following processing methods,

accessed from the Processing tab of the PowerCube property sheet:

You can further optimize cube creation by specifying one of the following processing methods,

accessed from the Processing tab of the PowerCube property sheet:

❑ Auto-partition. Enables the Auto-Partition tab, where you can set the parameters for Transformer

to devise a partitioning scheme.

❑ Categories. Minimizes the number of categories in a cube, the default option for models.

Transformer adds only categories that are referenced in the source data or specifically designated

to be included.

❑ Data Passes. Optimizes the number of passes through the temporary working files during the

creation of a cube. This option is beneficial only if the more efficient alternative, auto-partition-

ing, is not used (that is, your model implements features not supported with auto-partitioning).

❑ Direct Create. Adds all categories in the model to the cube while the data sources are processed.

Records that don't generate new categories are then directly updated to the cube. This optimiz-

ation method also predates the more efficient alternative, auto-partitioning. It may be used in

models that are expected to generate few new categories, and where all categories are expected

to be added to the cube. It is not available for individual cubes within a cube group.

❑ Incremental Update. Selecting this option may speed up cube updates by adding only new data,

without recreating the whole cube. You can create data sources that only contain new data and

use these data sources to incrementally update a cube. If your increments are small, incremental

updates may reduce the time needed to add records to the cube.

Note: The model must remain synchronized with the cube that you are incrementally updating.

If the model changes, you must recreate the cube, but you can add new categories to an incre-

mentally updated cube without recreating it. We recommend that you recreate your cube peri-

odically, in its entirety, rather than always using incremental updates. Incremental updates can

improve cube creation time up to a point but, depending on the frequency and number of

updates, you may find that processing time increases after a while.

Step

● In the Processing tab (PowerCube property sheet), select any of the following, and click OK:

● Optimization. Use to choose how Transformer optimizes the creation of a cube. Auto-

Partition is the default. See "Steps to Perform Auto-Partitioning".

● This Cube is Incrementally Updated. Use to add only the most recent data to an increment-

ally updated cube, thereby shortening processing time. See "Update Your Cubes Increment-

ally" (p. 163).

● Cube Creation. Use to create the cube or cube group locally. Enabled is the default. If data

related to a particular cube is unchanged since the last update, shorten processing time by

selecting Disabled.

● Store This PowerCube Compressed. Use to compress the cube for more compact storage.



● Enable Crosstab Caching. Use to store summaries for the initial crosstab in the cube, which

shortens access time for report users.

Note: Depending on your PowerPlay edition, portions of this topic may not apply to you. For


Features and Editions section in the Introduction chapter.

Use Multiple Data SourcesBy using multiple data sources for your model, you can reduce the total volume of data processed,

thereby shortening processing times in Transformer. If used in combination with calculated columns,

multiple data sources can minimize or eliminate the need for creating database table joins in an

external data access tool. Using multiple data sources also enables measure allocation.

Models created from multiple data sources are supported in all editions of Transformer.

Example

Data for products, customers, and orders is stored in a set of tables. If you provide this data from

a single source, you need the tables for Product, Customer, Custsite, Order, and Ordrdtl. However,

the source would contain many duplicate values and the joins between the tables would be relatively

complex.

Instead, you create three separate data sources: Products, Customer/Site, and Order/Order Detail.

The volume of data contained in each is less than that in the single source, and the joins between

Customers and Custsite, and between Order and Ordrdtl, are simple.

Steps

1. Using a data access tool, create each of the data sources required for your model.

Each source must contain sufficient data to establish context within the dimension map. You

cannot perform database table joins in Transformer.

2. Use the New Model wizard to add the largest structural data source to your model.

3. Use Insert Data Source (Edit menu) to add the additional sources to the Data Sources list.

Consolidate Data in PowerCubesConsolidation summarizes measure values during cube creation in Transformer, so that fewer

summarizing calculations need be performed in PowerPlay. Consolidation reduces cube size and

using the currently implemented auto-partitioning algorithm, does not increase cube creation times.

For more information, see the consolidate data example.

Transformer automatically consolidates measure values in cubes. For a list of instances, see "Con-

solidate Data: Overview " (p. 120).



The consolidation process is run against every source file in your model. Ensure that your temporary

file space is at least as large as the sum of all your source files.

Steps

1. Open the PowerCube property sheet for the cube that you want to consolidate, and then click

the General tab.

2. In the Consolidate box, select the appropriate entry, and then click OK.

Refresh Category InformationFor large, complex models, you may improve processing in Transformer by disabling the options

that refresh the category information. If this information subsequently changes in your source data,

remember to enable these refresh options again, before regenerating your categories.

Steps

1. Open the appropriate Level property sheet so that you can disable its refresh options.

2. Clear theRefresh check boxes for Label Column,Description Column, and Short Name Column.

3. Click OK.

Devising an Optimal Partitioning StrategyThe partitioning process passes through the data, often many times, consolidating and counting

the records in the temporary work files. Data is then divided into sets (transparent nested partitions).

One summary partition contains presummarized values from each of the lower-level detail partitions.

The categories that form the roots of the partitions are assigned partition numbers to indicate a

partition level.

Partitioning is useful when there are millions of rows of source data because it can significantly

reduce the time needed to access OLAP reports. There is a tradeoff, however: the more partitions

there are, the longer it takes to build the cube.

You can

● allow Transformer to automatically partition your cube. Let the built-in algorithm choose your

candidate partitioning levels, then consolidate records with sort (the default option).

● specify one or more dimensions for partitioning and use manual levels to add depth. This is a

good strategy if the dimensional hierarchies in your original model were too flat.

● check the log file to see if the number of rows in the summary partition or the number of records

and categories processed is decreasing from pass to pass. If it isn't, reduce the Maximum

Number of Passes setting on the Auto-Partition tab of the PowerCube property sheet.

● repeat with a larger partition size or manually fine-tune the partitions further

Evaluate Partitioning Approaches and Constraints

You can use two strategies to partition a cube successfully:



● Ensure that the majority of queries are answered from the first or upper partitions (called

summary partitions).

● Arrange partitions so that information required for most queries can be found within a single

partition.

When you develop a partitioning strategy, bear in mind of the following considerations:

● You can only partition on categories in the primary drill-down path. (However, auto-partitioning

may occasionally partition categories from dimensions that have alternate drill paths.)

● You cannot partition on root, leaf, or special categories.

● Avoid partitioning a dimension for which lowest-level detail reports are needed. Access times

are slower for these reports than for summary-level ones because data is returned from several

partitions.

● Reduce the number of records in each partition if the network is slow.

● Completely recreate a cube if you change the partitioning scheme.

Partitioning will not be used if any of the following applies:

● Consolidation is disabled for the model; that is, the Consolidate option on the General tab of

the PowerCube property sheet is set to No.

● The Optimization method is set to Categories (smallest-possible cubes) or Direct Update (fastest

build times).

● The cube uses externally rolled-up measures.

● The cube uses before-rollup calculated measures. Use calculated columns instead.

● Any cube in the model uses incremental update.

Additionally, the following environmental parameters change the range of partition sizes that you

can use and, in general, apply to all cubes:

● network bandwidth - Communication of information between a cube and the end user's com-

puter can involve multiple networks and host computers. If you use smaller partitions then you

can support a larger number of queries.

● cube production time frame - If you decrease partition sizes then query response time improves

because the volume of data required for query calculation is reduced. However, this increases

cube build time.

● acceptable query response time - Query response time is normally expressed in terms of an

average because it varies with the distance from the cube and with the complexity of the reports

executed. The number of simultaneous requests being processed and network capacity can also

affect query response time. You need to consider fluctuations in workload during the course

of a typical work day, in order to determine if the correct OLAP architecture and hardware

has been used.



● server and end user hardware - You should start capacity planning by estimating volumes of

incoming requests. Then you can design a system that has a combination of hardware capacity

and cube structure to support the request load. Large partitions are only practical on servers

that are capable of performing a large number of calculations per second.

Prepare Your Model for Partitioning

Partitioning-related performance improvements are most pronounced when models have hierarchical

dimensions and category-per-level ratios of 10:1 or less. Auto-partitioning does not work well on

models where the top level of a dimension has hundreds of categories and the leaf level directly

beneath it has thousands more.

If the dimensional hierarchies in your current model are too flat, try adding one or more manual

levels to bring the larger dimensions closer to the recommended 10:1 ratio.

You can do this by

● setting up clearly recognized, business-related groupings in your source data, such as commodity

types or customer sales channels, and importing them into your model as a new data source

column

● manually creating subdivisions such as geographic regions on the Transformer dimension map,

then using the category viewer to drag lower level categories to associate with the appropriate

parent levels

● adding calculated columns such as product number ranges or alphabetically sorted customer

groups

If these manually added categories are not needed in certain cubes or reports, you can create a

dimension or user class view and use the Suppress command to remove the unnecessary levels of

detail. However, if you suppress manual levels in a dimension view, this will reduce the effectiveness

of auto-partitioning.

Select the Best Partitioning Method

To speed up report retrieval times, accept the automatically set partitions in Transformer. These

pre-summarize the cube data into rows at one or more optimal partition levels.

For models that have relatively flat dimensional hierarchies, you may prefer to use manual parti-

tioning so that you can control how partitioning is done on the dimensions, levels and categories.

Other possible reasons for this choice include the following:

● You want to improve on the auto-partitioning strategy.

● You have large or unusually structured cubes.

● You want to tune performance for the most popular reports.

Define Partitions Automatically

To specify how auto-partitioning affects your cube, you can modify two parameters on the Auto-

Partition tab of the PowerCube property sheet:



● the desired partition size (based on estimated number of consolidated records). Transformer

selects groups of categories that satisfy the partition size criteria you specify and that optimize

query performance. Transformer always stops partitioning when the number of records in the

summary partition drops below the specified partition size.

● the maximum number of partitioning passes. Transformer continues for this number of passes

in an attempt to optimize query performance. It may not use all the passes that you specify,

but it won't use more. It uses one pass for each partition level that it creates.

As you decrease the desired partition size and increase the number of passes, the number of partition

levels created increases, and cube creation time increases. The last pass that Transformer performs

consolidates the summary partition. If an additional pass does not yield a reduction in summary

partition size then this pass can be skipped.

Transformer may read the source data several times to identify a good partitioning strategy, so

expect auto-partitioning to extend cube creation times, at least initially.

Auto-Partitioning Tips

● Experiment with different partition sizes, beginning with large partition sizes and working

toward smaller ones. Set the maximum number of passes to a large number while testing.

● Later passes may not produce significant improvements in query response time. If this happens,

you can decrease the number of passes to achieve an appropriate balance between query per-

formance and cube build time.

● If both the summary partition and the level-one partition have the same number of records

then the summary partition has not been consolidated. In this case, you should increase the

number of passes.

● If you have existing partitions that are manually defined, we recommend that you clear them

and let Transformer define the partitions automatically. You can then compare the two. Select

PowerCube Partition Status (Tools menu), and see how using auto-partitioning has optimized

your model. Choose the best solution.

Steps to Perform Auto-Partitioning

1. Open the property sheet for every cube in the model, click the Processing tab and ensure that

Optimization is set to the default (Auto-Partition).

2. Click the Auto-Partition tab and enter the following partitioning guidelines:

● specify the Estimated Number of Consolidated Records

● adjust the slider in the direction of Faster Cube Creation or Faster Cube Access

● specify a Desired Partition Size (start large, not more than 250, 000 records, then reduce

in incremental steps; test and repeat until runtime processing is acceptable)

● specify the Maximum Number of Passes (start with 5, or one pass for every manually

defined partition level)

3. In the Dimensions tab, select the dimensions that you want to include in your cube.



4. Click OK.

5. To view partitioning information, right-click a cube in the PowerCubes list and click PowerCube

Partition Status.

6. To remove partitioning, select Reset Partitions (Tools menu).

Define Partitions Manually

If your cubes are very large or unusually structured, you may achieve faster build times and better

runtime performance by manually defining your partitions. You do this by assigning partition level

numbers to categories in selected dimensions, preferably ones that are several levels deep, with

similar category-to-level ratios throughout.

Most models have dimensions in which the lowest-level details are accessed frequently, such as

product codes. In these dimensions, it is important to manage categories with a high parent-to-child

ratio and partition them accordingly. Too many partition levels will adversely affect lowest-level

detail reports.

Dimensions that are summarized frequently are also candidates for manual partitioning. Generally,

if 80 percent of your users create summary reports with categories from a given dimension, consider

partitioning at a high level in that dimension.

Before partitioning manually, we recommend that you run auto-partition, examine the log file, and

take note of the partition points selected by the auto-partitioning algorithm. Once you manually

assign partition level numbers, auto-partitioning is disabled and only categories and levels specified

manually get partitioned.

Using a small partition size for dimensions that are summarized frequently may contradict partition

size choices you made earlier. This is especially true if your cubes have frequently summarized

dimensions as well as dimensions with categories that are mainly used to report lowest-level details.

You can't use different partition sizes on different parts of the cube. You must decide whether to

favor summary reports or lowest-level detail reports.

To help you test your partitioning strategy, you can use the following Tools menu options:

● PowerCube Partition Status. Logs the record distribution resulting from your trial partition.

● Reset Partitions. Removes partitioning from your model, so you can test an alternative strategy.

Example

In a sales analysis model that contains time, region, and product dimensions, partitions are defined

based on product lines and sales regions. Separate partitions are defined for the Outdoor Products

line and the Europe region because

● most of your products are in the Outdoor Products line and most of your offices are in Europe,

so this is the most frequently queried information at runtime

● the consolidated summary partition for all Outdoor Product sales contains less than 500,000

records

● the consolidated summary partition for all European sales contains less than 500,000 records



Given this sales distribution, partitioning on both the regions and product line dimensions creates

the most useful summary partitions. The resulting consolidated records are readily accessed because

there is no waiting for rollups to be computed in the reporting component.

Manual Partitioning Tips

● A partition level is a set of categories that all receive the same partition number. These categories

can come from more than one level in a dimension. However, we recommend that you select

them from the same level and include a specific set of categories across the dimension. If more

than three levels of partitioning are required to achieve acceptable performance, consider adding

partitioning to another dimension or level.

● Choose dimensions that contain a large number of categories. Consider dimensions that contain

large numbers of levels compared to other dimensions in the model. Most often, these dimensions

offer the greatest possibility for consolidating rows during cube creation.

● Avoid partitioning dimensions that are frequently updated with new categories, or which contain

alternate drill-down structures, leaf-level subdimensions, or special categories. Do not specify

partition numbers for leaf categories, drill categories, or the main root category.

● When you manually define partitions, you must still specify the Maximum Number of Passes

setting on the Auto-Partition tab (Cube property sheet). The number of passes must be set to

the number of manually assigned partition levels, or to a number greater than that.

● You can include categories from different levels in the same partition level. However, a single

partition level cannot include categories from more than one dimension and you cannot partition

a suppressed category.

Steps to Partition Manually

1. Begin by calculating the required number of partitions, based on the following formula:

Number of partitions = (Number of source rows)/(Desired partition size)

Note: In most production environments, partition levels should not exceed 3, and partition

sizes should not exceed 500,000 records. Source rows are based on the source file.

2. Open the diagram for the dimension in which you want to define partitions, and choose a level

that contains approximately the number of categories identified in Step 1.

3. Open the property sheet for the category or level whose partition number you want to set.

4. In the Partition box, type the partition number you want to assign to the category or level and

click OK.

5. Assign the same number to each category in a partition level, starting with the lowest level and

working up to the top level in the dimension. Make sure each partition level number is equal

to, or one greater than, the partition number of its ancestor. If there is a gap in the numerical

order of the partitions, Transformer alerts you with an error message when you create the cube.

Note: Manually-specified partition level numbers don't necessarily define the order in which

partitions are placed in the cube. Partition level numbers may be reordered by Transformer.



6. When you are finished, click the Auto-Partition tab (PowerCube property sheet) and ensure

that you specify aMaximumNumber of Passes that is at least as large as the number of partitions

you calculated in step 1, and added in steps 2 to 5.

7. Create the cube and use the PowerCube Partition Status command (Tools menu) to view the

distribution of records:

● If no partitions are larger than your desired partition size, partitioning is complete.

● If some partitions are larger than your desired partition size, proceed to Step 8 and test

performance (access times) in your reporting component.

8. In your reporting component, drill down to the partition with the largest number of records.

If the performance is unacceptable, you may have to add another level of partitioning in this

dimension. Check the number of records in the summary partition. The goal is to reduce this

number.

9. If necessary, repeat this procedure for levels in other dimensions, ensuring there are enough

partition levels to yield the desired reporting performance. For each new level of partitioning,

remember to increase the partition number by 1.

Omit Dimensions

You can choose to omit dimensions even if you use auto-partitioning. Consider omitting a dimension

if it is used only for lowest-level detail reports.

If you omit detail dimensions, which are often very large, you can set a smaller partition size. Cube

build time will decrease and summary queries will be processed faster.

If the same model includes partitioning on detail dimensions, lowest-level queries are slower because

they report across multiple partitions.

Add New Partition Levels to Existing PowerCubes

You can add new categories to an existing partition level. For example, if a cube is partitioned on

a level called Regions, adding a new region to the level results in a new partition for that region.

You cannot add new partition levels to a model used to create an existing incrementally updated

PowerCube. To change partitioning for the model, you must repartition the model and rebuild your

PowerCube from scratch, using all the data from all increments.

Create Multifile PowerCubesBy default, a cube is automatically output to multiple files when it contains more than 30 million

data records, or exceeds 2 GB (formerly, the size limit for a single cube). You can change this

threshold by adjusting the MultiFileCubeThreshold setting in the Transformer .ini file. For more

information, see the Trnsfrmr.ini file topic.

Transformer determines the number of output files needed, taking the number of data records in

the cube, dividing by the threshold, and rounding up.



If the cube is partitioned, the contents are evenly distributed across the required number of multidi-

mensional partition (.mdp) files, and an additional multidimensional cube (.mdc) file is added to

hold the PowerCube metadata.



Chapter 7: Use Transformer in a Client-ServerEnvironment

Running Transformer in a client-server environment differs from running it from a personal com-

puter:

● You must define the connection to the server (done during installation).

● You must install the PowerGrid network daemon on the server. You can use NetInfo to ensure

it is running.

● You can create a prototype model on the client, then specify On the Server as the location where

the cube will be processed and upload the model to the server. You can check the progress of

your job by selecting Server Status (Server menu).

● Your client and server models must be synchronized.

For information on running Transformer in a client-server environment, see Transformer for UNIX.

Notes

● Transformer on UNIX only recognizes a limited set of data sources, including .iqd files,

delimited or fixed-field text files, and Cognos PowerHouse subfiles.

● You can use the concurrent processing ability of the server to quickly create many cubes from

different models.

● You can synchronize several models at once using the Transformer Model Definition Language

(MDL). For more information, see the online book Transformer MDL Reference.

● If your model is in MDL format, you can use an FTP program to copy it from your client

computer to the server.

● If the server or user name changes, click Maintenance (Server menu), click Server Setup, and

then define the new server configuration.






Chapter 7: Use Transformer in a Client-Server Environment

Chapter 8: Use Architect with Transformer

You must install Transformer on the computer on which you installed Architect. You will then be

able to use models and packages you created in Architect in Transformer.

Architect models (.cem) are created in Architect. Each model can contain one or more packages.

Each package contains

● all the information about a subset of a single business model, including business rules and cal-

culations

● business components, such as filters, prompts, and display formats or styles

● information about the data, such as label names (descriptive metadata)

There are differences in the way Transformer displays information when the data source is an

Architect package rather than a non-Architect data source.

Use an Architect Package as a Transformer Data SourceYou can create a Transformer model based entirely on an Architect package or you can use one

Architect package and a variety of non-Architect data sources. However, you can only use one

Architect package as a data source in each Transformer model.

Create a Transformer Model from an Architect PackageWhen you first create a model using the New Model wizard, you can only specify one data source:

either an Architect package, a single non-Architect data source, or the first in a series of multiple

data sources. In the latter case, the first data source you specify should be a large structural data

source containing columns that define the key dimensions, and hence the structure, of your model.

Once you have created a preliminary model, you can add other, non-Architect data sources to your

Transformer model.

Tips

● If you normally work in a UNIX server environment, you should create a prototype model

locally, to start. To speed up the design and test phase, use only a portion of your source file.

For more information, see Transformer for UNIX.

● If your model is based on more than one data source, you may want to allocate measures to

other levels, based on a constant or another measure in your model. Allocation between data

sources is supported by all editions of PowerPlay and Transformer.

Steps

1. From the File menu, click New to open the New Model wizard.


2. In the Model Name box, enter a name for your model.

Note: All models based on Architect packages use security. You do not need to select Include

Security in This Model.

3. Click Next and enter the name of your initial data source in the Data Source Name box.

4. Select Architect Model from the Data Source Type list, and click Next.

5. In the Access Manager Namespace box, select the namespace you want to use in your model,

and click Next.

6. In the User Class dialog, select a user class, and click OK.

7. Select the Architect model and package, and click Next.

8. On the last page of the New Model wizard, choose one of the following options:

● Use the Wizard to scan the metadata and create a preliminary model that you can edit. The

Candidate Hierarchies pane and the Common Measures pane are displayed, showing all

the candidate hierarchies and the common measures in the package that you are using.

Select the hierarchies, levels, and measures that you want and click Finish to create the

preliminary model.

● Exit the Wizard and use the Metadata Explorer to scan the metadata. When you exit the

Wizard the first time, the Metadata Explorer displays the package view pane with the

objects, attributes, and measures from the package, the Candidate Hierarchies pane with

the dimensions and levels, and the Common Measures pane. You can use the Metadata

Explorer options and controls to select the measures and candidate hierarchies that you

want to include in your model. Once you have made your selections, click OK to create

the model. You can use the Candidate Hierarchy button to show or hide the Candidate

Hierarchies pane. The next time you use the Metadata Explorer, it will appear in the view

it last displayed in.

Add an Architect Package to an Existing Transformer ModelIf you are working with an existing model, you can use the Insert Data Source command to open

the New Data Source wizard, and add an Architect package, or any other sources required, to your

model. Ensure that each data source has a unique name.

Steps

1. Right-click in the Data Sources list of your model and select Insert Data Source.

2. In the New Data Source wizard, enter the name of the data source in the Data Source Name

box.

3. Select Architect Model from the Data Source Type list, and click Next.

4. In the Access Manager Namespace box, select the namespace that was defined in the Access

Manager Configuration Manager, and click Next.



5. In the IBM Cognos Common Logon dialog, enter the userid and password corresponding to a

valid user in the namespace you selected, and click Log On.

6. Select the Architect model and packages that are available for your model, and click Next.

Transformer adds your specified source to the existing model, and you see your new source in

the Data Sources list.

7. Select Exit to the Wizard or Use the Wizard, and click Finish.

Metadata Explorer OverviewThe Metadata Explorer in Transformer shows all the objects that are included in any Architect

package: subjects, attributes, reference information, candidate dimensions, levels, and measures for

your model. You can select a component by clicking the check box beside it. The component will

appear in your dimension map when you click Apply or OK.

You can use the check boxes and buttons in the Metadata Explorer to design your model with a

degree of precision not available in models designed using the Wizard.

For information about the correspondence between Architect attribute usage settings and their roles

in Transformer models, see the Model with Architect Guide.

Models created by using Architect packages, like those using sources other than Architect, can

include

● a time dimension that contains the relevant divisions in the calendar or fiscal year, such as

quarter, month, week, and day, generally derived from a single column in your data source

● hierarchies (dimensions) with one or more drill-down paths. These hierarchies support analysis

of transactional values, such as quantity sold, by attributes, such as product type or vendor.

● manual levels added in the dimension map, to enhance the analytical value or to support

measure allocation

The left pane (package view) of the Metadata Explorer consists of a Measure tab that is selected

by default, and a Level tab. The Measure tab shows measures (performance indicators), such as

number of items shipped, from the Architect source. The Level tab shows the non-measure data

that will be used to create dimensions. Each tab offers two options:

● Subject, which displays all the subjects

● All, which displays all the attributes.

When you select one or more attributes in the Measure tab, the Candidate Hierarchies pane (top

right) shows possible Transformer dimensions and levels for that selection. In the Candidate Hier-

archies pane, you can select a whole hierarchy or the check box beside a level to include it in your

Transformer model.

The Common Measures pane (bottom right) shows candidate measures for your Transformer model.

You can select the check box beside a measure in the Common Measures pane to include it in your

model. If you click the Show Scope button, levels that are directly associated (in scope) with the

selected measure are highlighted in yellow.



Once you have made your selections, the Metadata Explorer builds and displays the candidate

hierarchies. Besides the measures you have selected, the Metadata Explorer displays other measures

that it finds in other levels in the hierarchy, depending on how you have set the Union and Intersect

controls. Select the Union button if you want the Metadata Explorer to display any dimensions

that are related to at least two measures that you have selected. Select the Intersect button if you

want the Metadata Explorer to display any dimension that is common to the selected measures.

In the Level tab, you can select a subject or attribute. When you make a selection, you will see the

proposed dimension for the subject displayed in the Candidate Hierarchies pane. In the Common

Measures pane, you will see the measures that relate to the subject. If you click the Show Scope

button, levels that are directly associated (in scope) with the selected measure are highlighted in

yellow.

In addition to the Show Subject/Show All options and the Union and Intersect controls, you can

use the Show Scope and Show Model controls to investigate the scope of your Architect source

data. The Show Scope button shows all levels associated with a measure you select from the Common

Measures pane of the Metadata Explorer. The Show Model button shows the levels and measures

already inserted in your Transformer model, as shown in the dimension map. You can use Show

Model to identify new candidates.

You can create alternate drill-down paths manually in Transformer, based on Architect packages,

by dragging the information from the package view pane to the dimension map.

You can drag attributes from the package view pane to the dimension map in the Candidate Hier-

archies pane to create whole dimensions, partial dimensions, or levels. To create a dimension, drag

the item to the dimension line. The new dimension creates a single level with the same name as the

dimension. To add levels, drag items from the package view pane to the levels area in the Candidate

Hierarchies window. You can also drag measures from the package view pane to the Measures

pane.

Design Your Transformer Model Using the Metadata ExplorerBy using the Metadata Explorer, you can analyze the entity-attribute relationships in your Architect

source data, select items for your model based on a candidate list of dimensions, levels, and measures,

and then see the scope of your choices.

You use the Union, Intersect, and Show Scope options to design your Transformer model in one

window, without having to run separate commands or jump to a different dimension view to see

the results of your choices.

Steps

1. If the Metadata Explorer was not automatically started from the New Model wizard, in the

toolbar, click Metadata Explorer.

2. In the Level tab (package view pane), select a subject from the package view pane, and review

its proposed dimensions and levels in the Candidate Hierarchies pane and the associated

measures in the Common Measures pane.

3. If you want to include the dimension in your Transformer model, select the check box beside

the level in the Candidate Hierarchies pane.



4. Repeat Steps 2 and 3 for each item in the Level tab and click Apply any time you want to insert

your choices in your Transformer model before continuing.

5. Click the Measure tab and, with the Show Scope and Show Model buttons selected, click a

subject in the package view pane.

6. Verify that the measure is associated with one or more of the levels you previously selected in

the Candidate Hierarchies pane.

Note: Yellow highlighting indicates a direct association.

7. If the measure is a performance indicator that you want to include in this model, select the

check box beside the item in the Common Measures pane.

8. Repeat Steps 5 to 7 for each item in the Measure tab.

9. When you are finished, click OK to close the Metadata Explorer. Transformer then builds the

required queries, based on your Architect source.

10. Confirm the results in the Data Sources list, then use the pop-up menu, dimension map, or

diagram to make further adjustments to your model, as required.

Filter Your Architect Data Source in TransformerYou can create different types of filters with different subjects in your Architect model and in

Transformer. Each type has different implications for your Transformer data sources.

● Filter an Entity in the Business Layer in Architect

Add a filter to an entity in the business layer if you want the filter to be included in every

package where that entity is used as a subject. Transformer will always apply the filter and it

will automatically become part of the SQL definition for your Transformer data source.

Use this filter when you want a filter that will be applied to every associated subject in every

package.

● Filter an Entity or an Entity Attribute for a User Class in Architect

Add a filter to an entity or to an entity attribute for a user class, if you want the filter to be

associated with a user class that will be available in the Transformer model if the modeler is a

member of the user class where the filter was included. Once it is applied, the filter will auto-

matically become part of the SQL definition for your Transformer data source.

Use this filter when you want the filter to be applied based on different user classes using the

entity or the entity attribute.

● Filter a Subject for a Specific Package or Without Specifying a Package in Architect

You can add a filter to a package in the package layer without specifically adding it to any

subjects or you can add a subject filter to a subject in a package. This filter becomes available

for Transformer to use, but will not be used automatically. To use it, you must import the filter

for each data source that you want it applied to. Once the filter is applied, it becomes part of

the SQL definition of your data source.



This is a reusable filter that you can set up in an Architect model as a productivity aid and

convenience when you are creating Transformer models. In Transformer, you can choose

whether you want to use this filter.

If you are creating filters that affect an attribute and you are importing the filters to the model

manually, ensure that you add them to every data source that contains attributes that are refer-

enced by the filters.

You can apply multiple filters against your Transformer model. Transformer combines the filters

using the AND operator.

Filters added to structural data sources only affect category generation.

Filters added to a transaction data source affect records in your cube.

Steps to Use an Architect Subject Filter in Transformer

1. Double-click on a data source belonging to your Architect model in the Data Sources list to

open its property sheet, and click the Filters tab.

Note: This filter must be defined in your Architect model and available in the package.

2. Click Import to open the Import Filter window.

3. From either the package or from one of its subjects, select the filter that you want to apply

against your data source, and then click OK.

4. To remove a filter, click the filter in your filter list, and then click Remove. This filter will no

longer be available in Transformer. However, since it is an Architect filter, it can be reimported,

if necessary.

Steps to Create and Use a Transformer Filter in Transformer

1. Double-click on a data source belonging to your Architect model in the Data Sources list to

open its property sheet, and click the Filters tab.

Note: This filter is unique to Transformer. It has not been defined in the Architect model and

is not available in the package.

2. Click Add to open the Filter Definition window.

3. Define your SQL expression to meet your requirements, and then click OK.

4. To remove a filter, select the filter in your filter list, and then click Remove. Filters designed in

Transformer are no longer available once they are deleted from your filters list.

Compare the Show Scope Function in Architect and TransformerThe following table shows how the Show Scope function applies in Architect and in Transformer.



UseColorScope

The values for the measure are recorded in this

level, or recorded in a lower level and rolled up to

this level. By default, this level appears with yellow

shading.

Level derived directly

Shows all levels associated with the measure you

select from the Common Measures pane of the

Metadata Explorer.

The same color is used to show this scope type in

the Metadata Explorer.

The values for the measure are recorded in a higher

level, but allocated to this level. These values can

be allocated by a constant or in proportion to

another measure. By default, this level appears with

green shading. The same color is used for each type

of allocation.

Level with allocated meas-

ures

Because measure allocation can only be done in

models created from multiple data sources, this

scope type is not shown in the Metadata Explorer.

There are no values for the measure in this level.Level has no direct associ-

ation A level that has no measure in scope appears white

in both the dimension map and the Metadata

Explorer.

The level is not related to the source column by

name, but a lower level is related. Therefore, the

records in the data source can be matched unam-

biguously to categories in the level. By default, this

level appears with light yellow shading.

Level derived indirectly

No equivalent appears in the Metadata Explorer.

The level is not related to an attribute or source

column by name. Consequently, there is no match

between records in the data source and categories

in the level.

Level not derived

A level to which no attribute or source column is

related appears white in both the dimension map

and the Metadata Explorer.



UseColorScope

The level is referenced (it is related to the source

column by name), but not derived. The records in

the data source can't be matched unambiguously

to categories in the level, because more than one

category can have the same name in the level (a

uniqueness problem). By default, this level appears

with red shading.

Level derived from a data

source that has missing

columns

No equivalent appears in the Metadata Explorer.

Share Category (General Tab, Category and Special CategoryProperty Sheets)

Share Category specifies the Object Identifier and shows the Category Code of another category in

the same dimension (the share object), that is used as the base for share calculations.

This item has been retained for compatibility with Transformer versions 5.x, in which you could

set and apply a share object. To create this type of category for an Architect data source in Trans-

former version 6.6.1 and above, add a calculated attribute.

On the General Tab (Level property sheet), Share Category is called Share Object.



Chapter 9: Reference

Cognos.ini and <RenditionName>.ini FilesThe Cognos.ini and <RenditionName>.ini initialization files contain database connection information

and other initialization settings that ensure forward and backward compatibility with other com-

ponents or renditions of the IBM Cognos platform. In cases of overlap, settings in the <Rendition-

Name>.ini and Trnsfrmr.ini files override the corresponding settings in Cognos.ini.

In PowerPlay Series 7, the <RenditionName>.ini file has the format cern.ini.

Cognos.ini and <RenditionName>.ini files are created or updated when the user installs an IBM

Cognos product running in the Windows environment. The default location for the Cognos.ini file

is in the \cognos folder, at the same level as cern. By default, the <RenditionName>.ini file is located

in the installation_location\bin folder used for your IBM Cognos products.

Transformer uses the following sections of Cognos.ini or, in PowerPlay version 6.6 and above, its

generic equivalents, <RenditionName>.ini or <CompanyName>.ini.

[PowerPlay Transformer]

Specifies the preferences for Transformer, as set by the user interface. Transformer maintains this

section and recreates any missing or invalid entries.

This [PowerPlay Transformer] section is maintained in the Cognos.ini file for compatibility with

PowerPlay versions prior to 5.2. Transformer now stores preference settings in Trnsfrmr.ini. Each

setting in Trnsfrmr.ini overrides any corresponding setting in Cognos.ini.

[Cognos Locations]

Provides information on the location of components of the IBM Cognos platform.

[Services]

Provides Transformer with information required to create model files.

[Service - Transformer Server]

Identifies the Bourne-shell or C-shell scripts used to open Transformer for UNIX from the Windows

environment.

[Databases]

Stores database definitions so that other components of the IBM Cognos platform can use the source

tables. To ensure backward compatibility, Cognos.ini remains the default location for all database

connection information.


[Connection List]

Contains a list of connection names for client/server commands. These connection names appear

in the Edit Network Connections dialog box. Used by PowerGrid when connecting to Transformer

for UNIX.

In Transformer versions 6.6 and above, a utility installed with PowerGrid copies the connection

information to a new file called Cogconnect.ini. The utility runs only once, during product install-

ation so, if you need to change any of the information, you must edit the Cogconnect.ini file directly.

[Connection Types]

Provides Transformer with a list of DLLs that are used for server connections. These connection

types appear in the Modify Connection dialog box. Transformer uses only Windows

sockets-compliant TCP/IP transports.

[<Connection Type> Config]

Provides information on the particular connection type, where <Config> is a keyname in the

[Connection Types] section. Used by PowerGrid when connecting to Transformer for UNIX.

[Connection - <connection name>]

Provides Transformer with parameters on specific server connections, where <connection name>

is a keyname in the [Connection List] section. You can configure this information by using the

Modify Connection dialog box, and its two descendant dialog boxes: Connection Timeouts and

Connection TCP/IP Setup.

[PowerPlay]

Provides Transformer with local user and company information. Depending on the country version

of your Windows operating system, this information may be derived from the Windows system

registry instead of the Cognos.ini file.

[PowerPlay DataServer]

Provides settings related to Transformer and the way OLAP source values appear in PowerPlay

client reports.

For example, you can change the memory allocation setting for read/write cache sizes used by

Transformer during cube generation. This may be required if your models contain a large number

of categories. For more information about adjusting the cache size, see the PowerPlay Administrator's

Guide or the IBM Cognos Configuration Manager User Guide.

You can also change the zero suppression setting to control how zeros are reported in PowerPlay

clients. Perform the following steps in your PowerPlay for Windows or PowerPlay for Excel client

applications:

Steps

1. In PowerPlay for Windows, from the File menu, click Preferences, and then click the Suppress

tab. In PowerPlay for Excel, from the PowerPlay menu, click Options, and then click the Suppress

tab.



2. In the Suppression Options box, select Zero Values.

3. Click OK.

● Tip: For best performance, suppress only zero values, but not computed zeros (aggregations

or calculations whose result is zero.) To suppress only zero values, make the following entry

in your cern.ini file:

● DisableMaxZeroSuppress=1

If you want to suppress computed zeros and zero values (the default), make the following entry

in your cern.ini file:

DisableMaxZeroSuppress=0

Or to restore the default behavior, remove the DisableMaxZeroSuppress entry from your

cern.ini file.

For more information about suppressing zeros, see the online Help for your client or, for third-

party OLAP data sources, the OLAP Server Connection Guide.

[PowerPlay DataServer Size Limitations]

In previoius releases, this section included a setting to prevent insertion of a record if the insertion

caused the resulting cube to exceed the size limit for local cubes.

For more information about size limitations, see the MultiFileCube Threshold Setting section of

the "Trnsfrmr.ini File" (p. 190) topic.

[PowerPlay Server List]

Provides Transformer with a list of valid data servers.

[PowerPlay Server - <server>]

Provides Transformer with additional information for each type of PowerPlay data server, where

<server> is a keyname in the [PowerPlay Server List] section.

[PowerPlay Server - PPDSRemote]

Specifies a client port, in the range of 1025 to 5000 inclusive, which is then used for all communic-

ation between the PowerPlay Enterprise Server query processor and the Windows client. Use for

security reasons, when you want to control which port ID the remote driver binds to, rather than

letting the operating system supply any available number.

For example, to bind the remote driver to client port 5000, type

Port=5000

Valid port IDs are 1024 < value ≤ 5000.

The same port ID can be used for all Windows clients, as long as they are installed on di-fferent

computers. Note that, when connecting remotely, users can only start one instance of the PowerPlay

client per computer.



Trnsfrmr.ini FileThe Trnsfrmr.ini file stores the preference values that you specify by using the Preferences command.

It contains a single section, [PowerPlay Transformer], which Transformer maintains by checking

for missing or invalid entries. The Trnsfrmr.ini file is created or updated when you change settings

in the Preferences dialog box.

Note: In Transformer Version 7.3 and subsequent releases, the cern.ini file includes a new section

called [Transformer Locale Type]. By default, the LocaleType is set to 1, meaning that the operating

system determines how decimal separators, thousands separators, dates, and currency symbols

display. However, in Transformer Win32 editions only, you can use Configuration Manager to

change this setting to 0. Transformer will then use some of the display settings configured in cerloc-

ale.xml, thereby maintaining consistency across the entire product suite for a given locale. This

change has no effect on the settings used during cube-building. For more information, see "Dates

and Numeric Values Not Formatted as Expected" (p. 300) in the PowerPlay Troubleshooting section.

File Location

Transformer automatically finds this file by looking in the executable directory of your most recently

installed rendition of the product. By default, the file is located in the folder: \cognos\cern\bin. To

retain your settings from previous versions of Transformer, the setup program moves any existing

Trnsfrmr.ini file to this spot, from the Windows directory or the home directory of a multi-user

Windows system.

AutoPartition Setting

Use Trnsfrmr.ini to set the AutoPartitionOff environment variable. A setting of 1 ensures that auto-

partitioned cubes always build with the Categories optimization method. A setting of 2 ensures

that model files imported into the current version of Transformer switch to Auto-Partition, unless

one or more cubes in a model contain features that preclude use of this default cube optimization

method.

MultiFileCube Threshold Setting

You can use the multifile PowerCubes feature introduced in Transformer 6.6 to create local

PowerCubes. By default, the MultiFileCubeThreshold setting in the [PowerPlay Transformer] section

of Trnsfrmr.ini is set to zero. This disables multifile generation, so that only a single cube is generated.

However, you can change this threshold setting so that very large cubes are automatically ouput

to multiple files. For example,

● set the threshold to a large unscaled integer (such as 30,000,000), if your cube is still less than

2 GB

● set the threshold to a smaller number (such as 1,000,000) to test or use the multifile feature on

smaller cubes

Transformer determines the number of output files needed, taking the number of data records in

the cube, dividing by the threshold, and rounding up. Cube partitions are spread evenly across these

multidimensional partition (.mdp) files, and an additional multidimensional cube (.mdc) file is

added to hold the PowerCube metadata.



CenturyBreak Setting

Use Trnsfrmr.ini to specify a cut-off date that determines whether the two-digit year (YY) in a six-

digit date is a 20th century date or 21st century date. Transformer stores values below the cut-off

date as 21st century dates:

● The default setting is CenturyBreak=20, which signifies that the dates 00 to 19 are interpreted

as the years 2000 to 2019, and the dates 20 to 99 are interpreted as 1920 to 1999.

● You can change this default to any value from 0 to 99. For example, if you change the setting

to CenturyBreak=05, all YY values between 00 and 04 are interpreted as 21st Century dates,

and all YY values at or above the cut-off date are interpreted as 20th Century dates (that is,

1905 to 1999).

If your data includes overlapping dates in two centuries (for example, 1900 to 1999 and 2000 to

2020), you must change your source data to use the 4-digit year format (YYYY).

MDL Save Option Settings

Use Trnsfrmr.ini to specify the following manually maintained MDL save options:

● To set .mdl save actions to verb format instead of structured MDL format, type VerbOutput=1.

● To set .mdl save actions so they do not use object identifiers, type ObjectIdOutput=0.

Note

In PowerPlay versions prior to 5.2, all preference settings were stored and maintained in the

Cognos.ini file. For compatibility, entries in the [PowerPlay Transformer] section of Cognos.ini are

still read; however, settings in Trnsfrmr.ini override those in Cognos.ini.

Rendition.cfg FileIn PowerPlay version 6.6 and above, a Rendition.cfg file is added during the installation process.

It specifies a base rendition name and a base rendition company: the key information needed to

point to the correct registry subtree (specific to a particular version and language). The registry root

is no longer hard-coded into the product.

For example, if the Rendition.cfg file contains the entries Base Rendition Name=cer6 and Base

Rendition Company=Cognos, the system looks for the required initialization settings in the

<RenditionName>.ini and <CompanyName>.ini files. In this case, it finds cer6.ini in the installation

directory for the cer6 (Cognos Enterprise Reporting 6) rendition, and Cognos.ini (which includes

the information required to make the proper database connections) in the cognos folder.

The Log FileTransformer tracks activity during processing by writing messages to a log file. By default, the log

file has the same name as the model file, but with a .log extension.

Log files are ASCII text files consisting of header information followed by a series of messages.

Each message is tab-delimited to show the following columns:



4321Column:

Message TextObject IDMessage SeverityProcessing

Date/Time

Contains:

Start cube update.000002B1405/13/1999

14:23:06

Example:

Processing Date/Time

The processing date and time is a time-stamp showing when the message was written to the log

file.

Message Severity

The severity of each message falls under one of four categories:

1. Severe Error

This can be caused by a system administration limit on CPU, disk, file, or transaction resources;

or by a corrupted model, cube, or temporary file. A severe error yields corrupted or inconsistent

files.

2. Error

Occurs at the transaction level. An error causes a cube to be marked as invalid, but leaves the

cube in a consistent (though possibly incomplete) state.

3. Warning

Doesn't impede processing, but indicates a potential problem.

4. Information

Object ID

The ID of the object processed, in hex notation. See the Preferences dialog box (Titles tab) to

activate object IDs.

Message Text

The text of each message includes information about processing and timing. Messages marked

Timing are especially useful to analyze a series of processing events.

You can do this by importing the log file into any spreadsheet application as a tab and comma-

delimited file. Because messages containing timing information are formatted with a comma (,),

after import, the timing information appears as a separate column in the spreadsheet. You can then

use the automatic filtering capability of the spreadsheet application, to focus your analysis on this

aspect of the processing.

Notes

In the Preferences dialog box, you can specify the following for log files:

● the log file name (the default is the model name)



● the directory to which Transformer writes the file

● the degree of detail (severity level) logged

● the elapsed minutes between each logging entry

● whether entries are appended to the log file



Appendix A: Additional Reference

Creating Drill-Down PathsWhen you create an alternate drill-down path, you designate an additional way of viewing the data

in the selected level. When drilling down in PowerPlay, you can choose which of these paths to

take.

Alternate drill-down paths always begin at a root category and extend to the selected convergence

level. For a convergence level to be valid, it must have source values that map to unique categories.

Otherwise, there is no way of distinguishing between categories.

If you create an alternate drill-down path to a level that is not flagged as containing only unique

categories, you are prompted to declare that level unique. Responding yes to this prompt has the

effect of selecting the Unique check box in the property sheet (Source tab) of the convergence level.

Defining ScopeMeasure Scope Dimension Map

Use the measure scope dimension map to focus on the portions of the model where measures have

meaning. By default, measure scope levels appear in one of the following states.


The values for the measure are recorded in this level, or recorded in a lower level and rolled up to

this level. By default, this level appears with yellow shading.

Level with Allocated Measures

The values for the measure are recorded in a higher level, but allocated to this level. These values

can be allocated by a constant or in proportion to another measure. By default, this level appears

with green shading. The same color is used for each type of allocation.

Level Has No Direct Association

There are no values for the measure in this level.

Data Source Scope Dimension Map

Use the data source scope dimension map to focus on the dimensions and levels with which a data

source is associated. By default, data source scope levels appear in one of the following states.



The level is related to a source column by name. Consequently, the records in the data source can

be matched unambiguously to categories in the level. By default, this level appears with yellow

shading.

Level Derived Indirectly

The level is not related to the source column by name, but a lower level is related. Therefore, the

records in the data source can be matched unambiguously to categories in the level. By default, this

level appears with light yellow shading.

Level Not Derived

The level is not related to a source column by name. Consequently, there is no match between

records in the data source and categories in the level. A level to which no source column is related

appears white in both the scope map.

Level Derived from a Data Source With Missing Columns

The level is referenced (it is related to the source column by name) but not derived. The records in

the data source can't be matched unambiguously to categories in the level because more than one

category can have the same name in the level (a uniqueness problem). By default, this level appears

with red shading.

Tips

● To change the default color for a level state, use the Dimension Map tab (Preferences dialog

box).

● You can perform the same actions on a scope dimension map as on a regular dimension map.

Quotation Marks in DataThe MDL syntax reserves quotation marks, (') and ("), to delimit textual data. All text strings are

enclosed in either double quotes, for example, "text", or single quotes, for example, 'text'.

Syntax in MDLOriginal Text

"New York"New York

"VP's Statistics"VP's Statistics




'Other "No-name" Brands'Other "No-name" Brands

When a text string contains both single and double quotes, Transformer changes some quotation

marks to avoid ambiguity. All quotation marks in the original text are converted to single quotes

('), and the entire string is enclosed in double quotes (").


"Other's 'No-name' Brands"Other's "No-name" Brands

"'London' and London ' ('prime')""London" and London ' ("prime")

"Steel Swivel Castor 2 1/2', Black"Steel Swivel Castor 2 1/2", Black

Note: Using quotation marks to designate Imperial measurement is not recommended. If you use

quotation marks to designate inches and feet, then the double quote used to designate inches is

changed to a single quote for feet.

Client-Server Model SynchronizationSynchronization involves copying model changes from client to server and from server to client,

and updating several time-stamps and identification numbers stored in the client and server models.

When Transformer synchronizes client and server models, it performs these steps:

1. Changes in the client model are incorporated into the server model. Internal time-stamps and

identification numbers are updated.

2. Changes in the server model are incorporated into the client model. Internal time-stamps and

identification numbers are updated. This process is the counterpart of step 1.

In performing steps 1 and 2, Transformer may encounter conflicting changes in associated client

and server models. In such cases, Transformer overrides server changes in favor of changes

made to the client model.

3. Current date and relative time relationships are recalculated on the client and transmitted to

the server model.

4. All internal stamps and identification numbers in both the client and server models are updated

to show that they have been synchronized. This ensures that future changes to either model

will be recognized as updates in the associated model.

Note: If you manually edit the MDL description of a model and then process it on either the client

or the server, do not change any of the internal identification numbers associated with synchroniz-

ation. These values ensure that models and cubes remain synchronized as they are changed. Identi-

fication numbers involved in model synchronization generally contain the keywords "Synchro" (as

in "SynchroStamp"), "Cycle" (as in "UpdateCycle"), or "Stamp" (as in "ClientStamp").



When To Synchronize

When you have made changes to either a client or a server model independent of the associated

model (that is, by not using commands in the Server menu), you should synchronize the models.

For example, if you update a model on the server by running Transformer on Unix from the UNIX

command line or via a crontab entry, you should later synchronize the model with the associated

client model.

It is not necessary to synchronize the client model with the server model following each production

update on the server. Use the Synchronize command when you want to make changes to the client

model or when you want to review the current contents of the model. Transformer automatically

synchronizes client and server models when you issue any of the commands on the Server menu.

Server ProcessesWhen you issue a Generate Server Categories or Create Server PowerCubes command, Transformer

performs these steps:

1. Transformer on Windows and Transformer on Unix establish a connection.

A dialog box may appear in which to type the password for your server account.

2. The client and server models are synchronized.

3. Transformer on Unix is started to generate categories for the server model by using the associated

server sources.

4. For a cube creation command, the cubes or cube groups for the server model are created by

using the server-based data sources and the contents of the server model.

The way Transformer on Unix uses server-based data sources depends on the timing set for

each data source. See the Timing properties in the General tab (Data Source property sheet).

5. Any resulting changes are sent from the server model to the client, where they are incorporated

into the client model.

Warning and Error MessagesIf warning or error messages appear, then you can view the troubleshooting information for the

message. Click here (p. 198) for more information about specific messages.

Warnings are issued only when you use the Check Model command. Errors are issued both when

you use the Check Model command and when a model check is performed implicitly during category

generation or cube creation.

PowerCube Status FiltersYou can use a combination of the two drop-down lists to filter the PowerCubes list. The following

are valid settings for Cube Creation:



MeaningFilter Value

Lists all cubes whether Enabled or Disabled.Enabled/Disabled

Lists all cubes that are set to be created during the cube

creation process.

Enabled

Lists all cubes that are set to be skipped during the cube

creation process.

Disabled

Valid settings for Status are:

MeaningFilter Value

Lists all cubes associated with the model, regardless of

status. The current status of each cube is shown in the

Processing tab of the PowerCube property sheet. For a

description of each cube status, see the What's This? Help.

Any Status

Lists all cubes that Transformer failed to create. Possible

states are Invalid, Failed, or Missing.

Errors

Lists all cubes from which warnings were detected during

the last creation or update.

Warnings

Lists all cubes that Transformer created successfully without

errors or warnings. The status is OK.

Successful

Data Source TypesThe following table outlines the data sources supported by Transformer. To view a complete list

of environments currently supported by IBM Cognos products, such as operating systems, patches,

browsers, Web servers, directory servers, and database servers, visit IBM Cognos on the Web (http:

//www-01.ibm.com/software/data/cognos/).





DescriptionSetting

The source table is described in an Impromptu Query

Definition (.iqd file). Input values are retrieved from a

supported database by executing an SQL query that is

stored in the .iqd file.

Impromptu Query Definition (valid

on client or server)

The contents of the .iqd file are stored in the model and the

embedded .iqd contents are refreshed when the data source

is opened (that is, when you generate categories or create

cubes), or when you click OK on the Data Source property

sheet.

For PowerPlay users to drill through to detail values using

Impromptu or Impromptu Web Reports, include an associ-

ated Impromptu report (.imr) file.

Note: Transformer does not support Impromptu HotFiles

as input.

Input values are found in a text file with one record per

line. The values in the first line represent column names.

The fields (column values) are delimited by the character

specified in the Field Delimiter box.

Delimited-field Text with Column

Titles (valid on client or server)


line. The fields are delimited by the character specified in

the Field Delimiter box.

Delimited-Field Text (valid on client

or server)

Input values are found in a dBase file.dBase Table (valid on client only)

Input values are found in an Microsoft Access file. Secured

(password-protected) files are not supported.

Access Table (valid on client only)

The data source table is described in a Microsoft Access

Query (.mdb file). Input values are retrieved from an

ODBC-supported database by executing an SQL query that

is stored in the .mdb file. Secured (password-protected) files

are not supported.

Access Query (valid on client only)

Input values are found in a Paradox file .Paradox Table (valid on client only)

Input values are found in a Lotus 1-2-3 crosstab file.Lotus 1-2-3 Crosstab (valid on client

only)

Input values are found in an Excel crosstab file.Excel Crosstab (valid on client only)



DescriptionSetting

Input values are found in a Lotus 1-2-3 spreadsheet data-

base file.

Lotus 1-2-3 Database (valid on client

only)

Input values are found in a Microsoft Excel spreadsheet

database file.

Excel Database (valid on client only)

Input values are found in a Cognos PowerHouse portable

subfile (.ps).

PowerHouse Portable Subfile (valid

on client or server)

Input values are found in a FoxPro file.FoxPro Table (valid on client only)

Input values are found in a Clipper file.Clipper Table (valid on client only)


line. Each field starts at the byte immediately following the

preceding field; the width of each field occupies a specified

number of bytes. Each row ends with a text line delimiter.

Fixed-Field Text (valid on client or

server)

Input values are found in a text file. Each field starts at the

byte immediately following the preceding field; the width

of each field occupies a specific number of bytes. The record

end is not marked by a text line delimiter.

Fixed Field and Record without CR

LF (valid on client or server)

Field DelimitersTransformer examines only the first byte in double- and multi-byte characters, to determine whether

a character in the input stream matches the delimiter character you specify. In addition, if the source

data is rendered in the DOS Code Page (OEM) character set, Transformer converts the delimiter

character to the OEM character set. Standard delimiters such as the comma, semicolon, or space

character remain the same across character sets. To use the tab character as a delimiter, type \t

(backslash t).

Character Sets

DescriptionSetting

See Windows (ANSI).Default



DescriptionSetting

The data source file was output by a DOS program, or by

an OS/2 program. It contains characters that map to the

active DOS code page.

DOS Code Page (OEM)

In the United States, most computers by default use Code

Page 437. For more information on code pages, see your

Microsoft Windows User's Guide.

This character set is also known as the IBM® PC character

set.

Use this character set to improve performance if the origin-

ating application does not use multi-byte characters.

Windows (ANSI) one-byte characters

The data source file was output by a Windows application

or a program on an ordinary ISO 8859-1 UNIX system.

For example, it may be a Comma-Separated Values file

(.csv) that was created by a Windows application such as

Microsoft Excel.

Windows (ANSI)

This character set is also known as the Windows Latin 1

character set. For more information on the Windows char-

acter set, see your Microsoft Windows User's Guide.

If you use a client-server model, you must use the same character set on both the client and the

server, or results with extended characters (above 128) will be unpredictable.

Isolation Levels in .IQD FilesYou can set isolation levels for Impromptu Query Definition (.iqd) files in the data source property

sheet.

DescriptionIsolation level

Uses the isolation level that was originally specified in

Impromptu when the .iqd file was created.

Default

Makes changes made by other transactions immediately

available to a transaction.

Read Uncommitted

Allows a transaction access only to rows that have been

committed by other transactions.

Read Committed

Prohibits other transactions from updating the row on

which a transaction is positioned.

Cursor Stability



DescriptionIsolation level

Ensures that rows selected or updated by a transaction will

not be changed by another transaction until the first trans-

action is complete.

Reproducible Read

Prohibits access by a transaction to rows inserted or deleted

since the start of the transaction.

Phantom Protection

Ensures that a set of transactions executed concurrently

produce the same result as if they were performed sequen-

tially.

Serializable

UNIX PathsWhen specifying a path to a UNIX host computer (server), be sure to use a fully qualified UNIX-style

path, with a forward slash character (/) as the directory delimiter. Paths can contain UNIX

environment variables if the FilenameVariable preference on the host computer is set to TRUE.

Date Input FormatsThere are two types of settings, specified in the locations noted below.

DescriptionSetting

As defined in the data source, includingPredefined

Clipper, dBase, or FoxPro

Impromptu Query Definitions, where the column has a data

type in the database

Paradox

PowerHouse portable subfiles, with a column marked as

date in the subfile dictionary

spreadsheets with date-formatted cells

Where the data source does not define the date format, such

as in text files, Transformer assumes the format to be

YYYYMMDD.



DescriptionSetting

As defined in the Regional Settings of your Windows Con-

trol Panel. The value is taken from the Short Date Style box

on the Date tab.

From Windows Control Panel

Note: This setting is available only on the General tab of

the Preferences dialog box (File menu.) When new columns

are created with this preference, the Date Input Format

setting in the Column property sheet initially reflects the

Windows Control Panel setting.

The settings below are available only in the Date Input Format box on the Time tab of the Column

property sheet. Use to set the date format that exactly matches your source data.

DescriptionSetting

Year, Month, Day; for example, 990413 or "1999-Apr-13"YMD

Day, Month, Year; for example, 130499 or "13 Apr 1999"DMY

Month, Day, Year; for example, 041399 or "Apr 13, 1999"MDY

Year, Month: for example, 199904 or 1999-AprYM

Transformer processes this setting as if the day component

was 01.

Month, Year; for example, 041999MY

Transformer processes this setting as if the day component

was 01.

Year; for example, 1999Y

Transformer processes this setting as if the month and day

components were 01.

Date Format CodesTransformer uses year, month, quarter, and day codes to construct date format codes. For levels

in a time dimension, you can combine the codes to construct date formats comprising years, quarters,

months, and days.

By combining codes, you can create date formats such as:

● YYYY MMM DD, which appears as 1998 Jan 01.

● YY "Q"Q, which appears as 98 Q1.



Note: For lunar years, quarters are labeled Q1-4, months are labeled 1-12 or 13, and days are

labeled 1-28.

Sample (English)MeaningCode

98A 2-digit year.YY

1998A 4-digit year.YYYY

1A 1-digit quarter indicator.Q

01A 2-digit month in the year.MM

JanThe abbreviated month name.MMM

JanuaryThe full month name.MMMM

01A 2-digit day.DD

SundayA day of the week.DDDD

1998/01/01, 1998-01-01Separator characters for other

date codes.

/, -, or space character

"(" shows an open parenthesis.The quoted string itself.Any quoted string

InclusionYou can prevent cubes from containing zero values for categories that lack data by including cat-

egories in the cubes only when they occur in the data source. Cubes that are more compact help

users to focus on the relevant data.

The Inclusion setting is available in various Category property sheets (Category, Drill Category,

Special Category), as well as the Level property sheet. In the Category property sheets, this setting

affects one selected category. You can specify a default Inclusion setting for all categories in a level

by setting the Inclusion property in the Level property sheet. The setting for the level applies to all

categories that use the default setting, and to all subsequently-created categories.

DescriptionSetting

Uses the Inclusion setting for the level in which the category

resides.

Default <<From Level>>



DescriptionSetting

Retains the category in the model and includes it in cubes,

even if it, or any of its descendants, fails to appear in the

data source.

Always Include

The category is also included if any of its descendants are

included and if the category has not been excluded, sum-

marized, or cloaked in a dimension view.

If necessary, Transformer includes ancestors of the category,

regardless of their Inclusion settings.

Note: Always Include is the default Inclusion setting when

you create a time dimension. If you exclude categories,

special categories that use relative time concepts (such as

Last Month) will not work properly.

Retains categories in the model, but excludes from cubes

those categories with a blank Source Value.

Suppress Blank Categories (levels

only)

Excludes the category from the cube but retains the cat-

egory's descendants and all data values associated with

them. The immediate descendants appear in PowerPlay at

the same level as the excluded category. The result is

identical to using the Suppress command (Diagram menu).

Suppress (categories only)

For a category to be excluded, the following conditions

must be true:

The category is not the share category of another.

The category is not referenced by a special category.

Retains the category in the model, but excludes it from

cubes when it and all its descendants fail to appear in the

current data source.

Include When Needed

If necessary, Transformer includes ancestors of the category,

regardless of their Inclusion property.

Excludes the category, its descendants, and all related data

from the cube. The result is identical to using the Exclude

command (Diagram menu).

Excluded (regular and calculated

categories only)

For a category to be excluded, the following conditions

must be true:

The category is not the share category of another.

The category is not referenced by a special category.



Allocation Types

DescriptionSetting

Uses the Allocation Type setting for the level in which the

category resides.

From Level (Category property sheet

only)

Suppresses allocation. PowerPlay users see "na" in the cat-

egory's descendants.

Do Not Allocate

Allocates the measure value associated with the current

category as a constant, to all descendants. PowerPlay users

see the constant value in all descendant categories.

Constant

Opens the Select a Measure dialog box. Use to allocate the

measure value proportionally to descendant categories by

using values from another measure as a weighting factor.

By Measure

For example, you might allocate the value of a fixed cost

measure to various regions based on another measure, such

as sales for each region. Select a measure that provides the

weighting values for the measure being allocated.

Partial and Extra WeeksPartial Weeks

If your time dimension contains a Week level in a calendar year, the day of the week that the year

ends on may not coincide with end of the week. (A lunar week always ends at the end of the week.)

For example, the last day of 2000 is a Sunday. By using the Partial Weeks option, you can specify

how to shift or split up the last week in the Week or the lowest date level in your time dimension.

The following example shows what would happen to a Sunday-to-Saturday week that spans

December 2000 and January 2001.

Dec. 2000/Jan. 2001

S M T W T F S

31 1 2 3 4 5 6

Example (Dec. 2000/Jan.2001)

DescriptionSetting

Period: Dec. 2000Places the week into the period in

which it begins.

First Period

S M T W T F S

31 1 2 3 4 5 6



Example (Dec. 2000/Jan.2001)

DescriptionSetting

Period: Jan. 2001Places the week into the period in

which it ends.

Last Period

S M T W T F S

31 1 2 3 4 5 6

Period Jan. 2001Places the week into the month

having the greatest number of days

in that week.

Largest Period

S M T W T F S

31 1 2 3 4 5 6

Period: Dec. 2000Creates two distinct partial weeks

by splitting the spanning week

between the two periods. On the

dimension diagram, both weeks are

displayed.

Always Split

S M T W T F S

31

Period: Jan. 2001

S M T W T F S

1 2 3 4 5 6

Period: Jan. 2001Splits the spanning week between

the two periods, provided that each

period contains at least two days.

Otherwise, places the week into the

period to which most of the days

belong.

Split > 1 Day

S M T W T F S

31 1 2 3 4 5 6

Extra Weeks

When you use lunar time periods, a year contains 52 weeks of 7 days each, yielding 364 days. This

is either one or two days less than the calendar year of 365 days or 366 days (for a leap year).

Eventually, after a few calendar years, the number of "remainder" days begins to accumulate.

Transformer can use these days to create an extra week category, and re-synchronize. This ensures

that the lunar year remains within the boundaries of the calendar year, plus or minus the Add an

Extra Week setting on the Drill Category property sheet.

Relative Time Settings

DescriptionSetting

Indicates that the special category is not a relative time

period. Use to cancel a previously established relative time

period without deleting the special category and its existing

substructures.

None



DescriptionSetting

The current time period.Current <Period>

For example, if the current date is 2000/05/30, then the

Current Month is 2000/05.

The period-to-date.<Period> To-Date

For example, if the current date is 2000/05/30, then

the Quarter To-Date category is a specially constructed

version of 2000 Q2 that contains only the months of April

and May (assuming a standard calendar year of four quar-

ters beginning in January).

the Year To-Date category is a specially constructed version

of 2000 that contains 2000 Q1, and a specially constructed

version of Q2 that contains only the months April and May.

the Life To-Date category contains all previous years plus

the Year To-Date categories.

A group of relative time periods consisting of a period-to-

date category and a period-to-date, prior period category.

<Period> To-Date Grouped

For example,

Quarter To-Date Grouped consists of a Quarter To-Date

category and a Quarter To-Date, Prior Quarter category.

Year To-Date Grouped consists of a Year To-Date category

and a Year To-Date, Prior Year category

The last period at this level.Last <Period>

For example, if the current date is 2000/05/30, then

the Last Month category is 2000/04.

the Last Quarter category is 2000 Q1 (assuming a standard

calendar year of four quarters beginning in January).

the Last Year category is 1999.



DescriptionSetting

The same period in the most recent higher-level period.Same <Period>, Prior <Higher-Level

Period> For example, the current date is 2000/05/30, which falls in

the second month of the second quarter of a standard

calendar year. Given these settings, then

the Same Month, Prior Quarter category is 2000/02, which

is the second month of the first quarter.

the Same Month, Prior Year category is 1999/05, which is

the fifth month of the previous year.

The periods-to-date in the context of the most recent higher-

level period.

<Period> To-Date, Prior <Period>

For example, the current date is 2000/05/30, which falls in

the second month of the second quarter of a standard

calendar year. Given these settings, then

the Quarter to Date, Prior Quarter category is a specially

constructed version of 2000 Q1 that contains only the first

and second months (January and February) of that quarter.

the Quarter to Date, Prior Year category is a specially

constructed version of 1999 Q2 that contains only the first

and second months (April and May) of that quarter.

A time period of your own choosing. When you select

Custom, you can change the Details associated with your

relative time categories.

Custom

Details

For any Relative Time selection other than Custom, the default detail settings cannot be changed.

However, they are useful for showing how a particular relative time category is constructed, or for

starting with a pre-defined selection, then changing the settings to create Custom periods such as

the following:

● 2000/03, which is two months before the month in which the current period occurs.

Current Period = 2000/05/30

Basic Approach = Single Category


Target Offset = -2

Context Period = <unspecified>

Context Offset = 0



● 1999/10, which is two quarters ago, one month before the month in which the current period

occurs in the current quarter.


Basic Approach = Single Category


Target Offset = -1

Context Period = Quarter

Context Offset = -2

● A category that spans all years up to 2000, as well the first quarter of 2000 and the months

April and May, 2000.


Basic Approach = Period To-Date Total

To-date Period = Life


Target Offset = 0


Context Offset = 0

● A category that spans the quarters 1999 Q1, 1999 Q2, 1999 Q3, and the months 1999/10 and

1999/11. This is Year to Date, two quarters ago, up to the same month in that quarter as the

current month is in the current quarter (in this case, the second month of the quarter).


Basic Approach = Period To-Date Total

To-date Period = Year


Target Offset = 0


Context Offset = -2

● A category that spans a five month period ending in July 1999. This is a five-month running

total, starting three quarters ago, where the ending month in that quarter is one month before

the position of the current month in the current quarter.


Basic Approach = N-Period Running Total

Number of Periods = 5


Target Offset = -1




Context Offset = -3

Degree of DetailThe Degree of Detail for a date column does not necessarily reflect the amount of detail in the

source data. It defines the level at which Transformer allows reporting to occur. For example, a

time column with values for Year, Quarter, and Month can be restricted to only allow quarterly

reports. You cannot specify a Degree of Detail that is lower than the lowest level indicated by the

Time Input Format on the Time tab (Column property sheet). The setting does not apply to regular

columns.

ToSelect

Have Transformer set the Degree of Detail for the date

column. Transformer bases the frequency on the lowest

level as defined by the Time Input Format setting.

Unspecified

Set a specific degree of detail for date-time values. Reporting

of measures in the time dimension is restricted to the level

of detail you select. If the level is not the lowest, the meas-

ures for levels below your selection become candidates for

allocation.

Any other item in the list

Level UniquenessWhat is Level Uniqueness?

Depending on your PowerPlay edition, portions of this topic may not apply to you. For more


and Editions topic in the Welcome chapter.

Level uniqueness in Transformer means that the level contains only one instance of each category

source value. In practical terms, this means no two categories in the same level can represent

● identically-named but distinct categories, as in cities with the same name in two or more states.

For example, in a Regions dimension with levels Country, State, and City, there might be cities

named Burlington in both Massachusetts and Vermont. Burlington is ambiguous without the

context of its ancestor levels, hence the level City is not unique.

● a "unique" real-world entity that occurs in different contexts, as in a single customer that

transacts business in several different cities. For example, in a Customers dimension with levels

Country, State, City, and Customer, a customer named Acme Metals might transact business

in Chicago, Philadelphia, and Orlando. In this case, the Acme Metals category will exist under

each of these cities. It is ambiguous without the context of its ancestor levels. Hence, the level

Customer is not unique.



When is Uniqueness Required in a Level?

1. Categories in a level must be unique when the level is a convergence level for alternate drill-

down paths.

2. Categories in a level must be unique when a model uses multiple data sources, where some

sources contain details about the structure of the dimensions, and other sources supply measure

values but lack details about dimension structure.

3. Categories in your time dimension must be unique if you are implementing currency conversion.

Why is Uniqueness Required in a Level?

1. In order to generate accurate categories for PowerPlay, Transformer must be able to identify

each category in the level by its source value alone, without reference to its ancestor category.

If a convergence level contains multiple occurrences of the same category value, Transformer

cannot unequivocally determine which drill-down path to take when generating categories.

For example, if Burlington appears in the City level, is it in the state of Vermont or Massachu-

setts? To make the distinction, you must associate the city with its state. For example, concat-

enate Burlington+Massachusetts and Burlington+Vermont (by using a calculated column) to

create unique categories, with different source values.

If you indicate that all categories in your convergence level are unique when, in fact, they are

not, invalid results occur when Transformer attempts to associate values with these categories.

Transformer places any ambiguous or non-unique source values into the first associated category

it finds.

2. It is common practice to have some data sources that contain transactional data, and other

data sources that contain only structural data. In star-schema relational databases, these are

referred to as fact tables and dimension tables respectively, and are used to significantly reduce

the amount of data imported into Transformer. If you create a dimension that uses columns

from multiple data sources, Transformer must be able to relate categories from structural

sources to categories from transactional sources.

For example, if a Regions dimension contains the levels Country, State, and City, data for

Country and State may come from a structural data source, while another (transactional) data

source supplies measure values at the City level. You can view this relationship by using the

Show Scope command. When generating categories, Transformer must be able to identify each

category at the City level by its source value alone, without reference to the State category.

When you indicate that the City level is unique, Transformer treats all categories at that level

as though they had a single source.

Date Functions

Creates categoriesThis setting

That use the value in the Source column as is.None



Creates categoriesThis setting

Based on the standard calendar in the format YYYY or YY.Calendar Year

Based on a lunar year, which contains exactly 52 weeks.Lunar Year

Based on the standard calendar in the form YYYY Q, or

YY Q, where Q is the Quarter number (1, 2, 3, or 4).

Calendar Quarter

Based on lunar quarters, which contain exactly 13 weeks.Lunar Quarter

Based on the standard calendar, in the format YYYY/MMM

or YY/MMM.

Calendar Month

Based on a lunar month, which contains exactly 4 weeks.Lunar Month

Based on repeating 3-month patterns containing 4 weeks,

4 weeks, and 5 weeks.

4-4-5 Week Months



4-5-4 Week Months



5-4-4 Week Month

Based on weeks, in the format YYYY/MMM/DD or

YY/MMM/DD.

Week

When weeks appear in a time dimension, you must specify

the way in which weeks split over month boundaries, and

the day that marks the start of the week. You can set these

properties in the Time tab (Drill Category property sheet).

In the form YYYY/MMM/DD or YY/MMM/DD.Day

Note: Because lunar years are 1 or 2 days shorter than calendar years, lunar periods (Lunar Year,

Lunar Quarter, Lunar Month, 4-4-5 Week Month, 4-5-4 Week Month, and 5-4-4 Week Month)

leave unassigned days that, over several years, comprise entire weeks. To manage the unassigned

days, you can set an "Add an Extra Week" rule on the Drill Category property sheet for the drill-

down path.



PowerCube Optimization Methods

DescriptionSetting

Minimizes the number of categories in a cube. Transformer

adds only categories that are referenced in the source data

or specifically designated to be included. Categories

optimization requires an extra data pass for each cube to

find the categories required for that cube.

Categories

Optimizes the number of passes through the temporary

working files during the creation of a cube. Transformer

assumes that all categories will be required in the resulting

cube, and does not pass through the source data to

determine which categories are required. Although included,

unreferenced categories will not be visible in PowerPlay.

Data Passes

Adds all categories in the model to the cube before the data

sources are processed. Records that don't generate new

categories are then directly updated to the cube.

Direct Create

This optimization method is best used with models which

are expected to generate few new categories, and where all

categories are expected to be added to the cube.

Note: This setting is not available for individual cubes in a

cube group.

Enables the Auto-Partition tab, where you can set the

parameters for Transformer to devise a partitioning scheme.

Auto-Partition

Note: This is the default optimization setting.

Rollup FunctionsThe rollup functions specify how measure values are evaluated in PowerPlay. There are three rollup

types:

● Regular Rollup is how PowerPlay or Transformer rolls up measure values from lower levels to

higher category levels. Transformer applies the rollup function when creating a cube; PowerPlay

applies them at runtime.

● Time State Rollup is how Transformer represents the state of a measure at specific times.

For example, if a model tracks the number of active customers at the end of each quarter, you

can set up a time state measure to report the number of customers active at a specific time. This

is more useful than the quarterly sum of the number of customers serviced during each month

in a quarter.



● Duplicates Rollup is how Transformer evaluates duplicate records in the source data.

Note: When consolidation is enabled, Transformer performs the duplicates rollup first and then

regular rollup. To enable consolidation, use the Consolidate box on the General tab (PowerCube

property sheet).

Regular Rollup

Results in PowerPlaySelection

Shows the sum of the values of all descendant categories of

the current category.

Sum

Shows the minimum data value among all descendant cat-

egories of the current category.

Minimum

Shows the maximum data value among all descendant cat-

egories of the current category.

Maximum

Shows the average of the values of all records of descendant

categories of the current category.

Average

Shows the number of records that contain non-null values

in all descendant categories of the current category.

Count

Shows the number of records (including those containing

null values for this measure) for all descendant categories

of the current category.

Count All

Shows the value 1, if any records for a descendant category

contain values.

Any

Shows the value 0, if no records exist for this measure or

all records that do exist have null values for this measure.

Shows source values that have been directly assigned to

specific data records.

External

Time State Rollup


Shows rollups not associated with a time state measure.

Use when the measure is not a time state measure.

None (Regular Rollup)

Shows the smallest measure value from all categories in the

time period being examined.

Minimum




Shows the largest measure value from all categories in the

time period being examined.

Maximum

Shows the measure value from the first subordinate period

in the time period being examined.

First Period

For example, if a time dimension contains Years, Quarters,

and Months, and you are examining data at the Quarter

level, PowerPlay shows (for each quarter) the measure value

from the first month of each quarter.

When you examine data at the Year level, PowerPlay shows

the first value from the first month in the first quarter of

each year.

Shows the measure value from the last subordinate period


Last Period

For example, if a time dimension contains Years, Quarters,

and Months, and you are examining data at the Quarter

level, PowerPlay shows (for each quarter) the measure value

from the last month of each quarter.

When you examine data at the Year level, PowerPlay shows

the value from the last month in the last quarter of each

year.

Shows the measure value from the category that is desig-

nated the "current period" in the time dimension. If the

time period being examined does not include the current

period, the result is identical to Last Period.

Current Period

For example, a time dimension contains Years, Quarters,

and Months, where Quarter 1 starts in January. The current

period is set to April 2000.

At the Year level, PowerPlay shows the measure value for

April 2000. At the Quarter level, PowerPlay shows the

measure value for April in Quarter 2 (because April is the

current period), but shows the value of the last active month

in every other quarter (that is, March for Quarter 1;

September for Quarter 3; December for Quarter 4.

Shows the average of the measure values from all categories


Average



Duplicates Rollup

Transformer ActionSelection

Uses the Regular Rollup function specified for the measure.

See the note below.

None (Regular Rollup)

Sums the measure values found in the duplicate records,

except for external rollups (overridden).

Sum

Takes the smallest of all values for the measure found in

the duplicate records.

Minimum

Takes the largest of all values for the measure found in the

duplicate records.

Maximum

Takes the average of all values for the measure found in

the duplicate records.

Average

Note: If Regular Rollup is set to Average, duplicates rollup

is performed before regular rollup.

Takes the first value of all values for the measure found in

the duplicate records, by using the order of the records in

the data source.

First

Takes the last value of all values for the measure found in

the duplicate records, by using the order of the records in

the data source.

Last

Note: A Duplicates Rollup setting other than None forces Transformer to consolidate the source

file when generating cubes, as described in the following table.

Transformer ActionConsolidation Setting

Sorts the source file, then consolidates duplicate records by

using the Duplicates Rollup setting.

Default or Yes (With Sort)

Overrides the Consolidation setting, sorts the source file,

then consolidates duplicate records by using the Duplicates

Rollup setting.

No

Consolidates the source data by using the Duplicates Rollup

setting without first sorting the source file.

Yes (Presort)



Expression EditorThe Calculation dialog box contains a common expression editor that is shared with other compon-

ents of the IBM Cognos platform. You select elements one by one from the set of available expres-

sions in the left pane of the dialog, and build your expression in the right pane. As you enter each

element, the syntax is evaluated against standard mathematical rules.

The available elements vary according to the type of calculation and the position in the expression,

and each is explained in a ToolTip window in the Calculation dialog box. Note that you can only

use components that appear in the Available Components pane of the Expression Editor.

For more information on the functions, operators, constants, and components provided by the

common expression editor, see the Expression Editor online help.

Search WildcardsYou can use the following wildcard characters in the Find in Category dialog box when the Use

Wildcards check box is selected:

ExampleFindsCharacter

"^inter" finds "interesting" and

"interfere", but not "splinters".

The beginning of a string.^

"$in" finds "in" and "within",

but not "interfere".

The end of a string.$

"to?" finds "top" and "ton",

but not "to".

Any single character (except

newline).

?

"files~" finds "file" and "files",

but not "filed".

Zero or one occurrence of the

preceding character (or sub-

expression).

~

"can*" finds "can" and

"Canada".

Zero or more occurrences of any

characters (except newline).

*

"filex#" finds "file" and

"filexxx".

Zero or more occurrences of the


expression).

#

"filex@" finds "filex" and

"filexxx", but not "file".

One or more occurrences of the


expression).

@



ExampleFindsCharacter

"localis|ze" finds either "local-

ise" or "localize".

Either the preceding character

(or sub-expression) or the follow-

ing one.

|

"p[iu]ck" finds "pick" and

"puck". "[a-d]ay" finds "bay"

and "day", but not "gay".

"m[!u]st" finds "mist", but not

"must". "m[!a-o]st" finds

"must" but not "most".

"Apr[-/]98" finds "Apr-98" or

"Apr/98"

Any character within the brack-

ets. Ranges of characters can be

specified by using a hyphen (a

hyphen at the start matches

itself). An exclamation point at

the beginning causes the set of

characters to be inverted; e.g.

[!a-m] matches everything except

a through m.

[ ]

"ab(cd)#e" finds "ab" followed

by zero or more "cd" combina-

tions followed by "e".

Sub-expressions, so that repeti-

tion and alternative wildcard

characters can be applied more

generally.

( )

"what\?" finds "what?".The next character literally.

Allows wildcard characters to

treated as normal characters.

\

Note: The following characters are reserved and should not be used: "<", ">", "{m,n}", and "\:x".

To find "<" or ">" literally, use "\<" or "\>".

Category Actions in Diagrams

SummarizeApexSuppressCloakExclude

Omit specific categor-

ies and their descend-

ants from all cubes cre-

ated from the model.

Omit specific categor-

ies and their descend-

ants from cubes created

by using dimension or

user class views.



SummarizeApexSuppressCloakExclude

Prevent specific categor-

ies from being accessed

by the user class (and

its descendant user

classes).

Eliminate ancestor cat-

egories from cubes cre-

ated by using the

dimension or user class

views.

Prevent ancestor cat-

egories from being

accessed by the user

class (and its descend-

ant user classes).

Show rollup values for

descendants of a cat-

egory without showing

the descendants them-

selves.

Enable members of a

user class (and its des-

cendant user classes) to

see rollup values for

descendants of a cat-

egory without seeing

the descendants them-

selves.

ExcludeEliminates categories and their descendants. Transformer shows categories and hides their descend-

ants. Users can drill down only to the parent of the excluded category.

You can apply exclude at the dimension level of the category viewer, in a dimension view, and

in a user class view.

Exercise caution when excluding categories from a dimension that contains alternate drill-down

paths. If you exclude categories from one path, the data is excluded from all drill-down paths. See

"Dimension Views in Alternate Drill-Down Paths " (p. 232).



CloakEliminates categories and their descendants, but rolls up their values for representation in the parent

category. Transformer shows categories and hides their descendants. Users can drill down only to

the parent of the cloaked category.

You can apply cloak in a dimension view and a user class view. You can remove an alternate

drill-down path from a cube by cloaking the associated drill category.

You cannot cloak special categories or categories in a convergence level of an alternate drill-down

path.

SuppressEliminates categories, maintaining links from the parent to the child categories. Transformer shows

categories. Users bypass the suppressed category when they drill down.

You can apply suppress at the dimension level of the category viewer, and in a dimension view

(but not in a user class view).

You cannot suppress root categories. In dimensions with alternate drill-down paths, Transformer

suppresses the drill category in the primary drill-down path by default. You can reverse this sup-

pression if you wish.

ApexMakes it the highest-level category. Transformer suppresses all ancestor categories and hides all

siblings (and the ancestors of all siblings) of the apex category. Users see only the apex category

and its descendants.

You can apply apex in a dimension view and a user class view.

You cannot apex special categories. In an apexed dimension, any special categories that refer to

regular visible categories become children of the apexed category. You can reverse apexing by re-

apexing the root category.


SummarizeEliminates descendants, but rolls up their values for representation in the selected category. Trans-

former shows categories. Users can drill down only to the summarized category. They cannot see

any of the category's descendants.

You can apply summarize in a dimension view and a user class view.

You cannot summarize special categories.

Union of User Class ViewsIn PowerPlay, if a user is a member of more than one user class, that user can open a cube as a

member of a single user class, or as a member of the combined user classes, using the <All User

Classes> option.



When opening a cube using the <All User Classes> option, PowerPlay shows the categories that

represent the combined access rights of all the user classes that the user is a member of. Anything

that users can see due to membership in any individual user class is visible to the union of user

classes. The advantage of this is that the user doesn't have to open the cube multiple times, as a

member of each user class, in order to gain access to all authorized data.

However, in certain situations, it may appear that PowerPlay is restricting users to less than what

they can see under their individual user classes. The apparent discrepancy is resolved if the user

understands how particular user class views are defined in the model. The type of view, such as

exclude, apex, or omit, affects the behavior, as does the application of that view, whether to a single

dimension or across multiple dimensions.

When you apply user class views to a dimension of a cube, you can limit access to specific categories

in the following ways:

● by Excluding categories to eliminate the categories and their descendants from view

● by Cloaking categories to eliminate the categories and their descendants from view, while

retaining the rollup values in the ancestors

● by Summarizing categories to eliminate only the descendants from view, while retaining the

rollup values

● byApexing a category to make it the highest-level category of the dimension, thereby eliminating

ancestors and siblings

● by Omitting an entire dimension so that its categories, data, and rollup values do not appear

Cubes that are created using user class views contain all the data defined in the model. However,

the categories that a particular user is able to see are determined at runtime by the PowerPlay engine.

Notes

● You cannot apex on a union with an alternate drill path.

● The following examples depict the union of two user classes in the Great Outdoors Company,

across the Products or Locations dimension. The effect of the union of more than two user

classes is cumulative. Evaluate the union of the first two user classes, then apply the next user

class to the result, and so on. The order of application doesn't matter; any order produces the

same result.

Scenario Involving ExcludeThis scenario involves the following user classes and applied views:

● UC1 - Exclude Environmental Line

● UC2 - Exclude Outdoor Products

● UC3 - Exclude Asia Pacific

● UC4 - Exclude Asia Pacific and Exclude Outdoor Products

Individually, UC1 cannot see Environmental Line but sees the following.



LocationsNorth AmericaEuropeAsia PacificUC1

880,453419,077337,514123,862Outdoor Products

302,216136,67293,26472,280GO Sport Line

1,182,669555,749430,778196,142Products

Individually, UC2 cannot see Outdoor Products but sees the following.


1,082,594514,911382,772184,911Environmental

Line

302,216136,67293,26472,280GO Sport Line

1,384,810651,583476,036257,191Products

Individually, UC3 cannot see Asia Pacific but sees the following.

LocationsNorth AmericaEuropeUC3

756,591419,077337,514Outdoor Products

897,683514,911382,772Environmental Line

229,936136,67293,264GO Sport Line

1,884,2101,070,660813,550Products

Individually, UC4 cannot see Asia Pacific or Outdoor Products, but sees the following.

LocationsNorth AmericaEuropeUC4

897,683514,911382,772Environmental Line

229,936136,67293,264GO Sport Line

1,127,619651,583476,036Products

If multiple restrictions are applied to a dimension, the following occurs:

● UC1+UC2: members of both, logged on as <All User Classes>, see all the data because it is

visible in one or more individual cases.



LocationsNorth AmericaEuropeAsia PacificUC1+UC2


1,082,594514,911382,772184,911Environmental

Line

302,216136,67293,26472,280GO Sport Line

2,265,2631,070,660813,550381,053Products

● UC2+UC3: members of both, logged on as <All User Classes>, cannot see data for Outdoor

Products in the Asia Pacific, because it is not available to either individual user class.


756,591419,077337,5140Outdoor Products

1,082,594514,911382,772184,911Environmental

Line

302,216136,67293,26472,280GO Sport Line

2,141,4011,070,660813,550257,191Products

● UC1+UC4: members of both, logged on as <All User Classes>, cannot see data for Environmental

Line in the Asia Pacific, because it is not available to either individual user class.



897,683514,911382,7720Environmental

Line

302,216136,67293,26472,280GO Sport Line

2,080,3521,070,660813,550196,142Products

Scenario Involving ApexThis scenario involves the following user classes and applied views:

● UC5 - Apex Environmental Line

● UC6 - Apex Outdoor Products



● UC7 - Apex Asia Pacific

● UC8 - Apex Asia Pacific and Apex Environmental Line

Individually, UC5 only sees Environmental Line. In the Dimension viewer, no other products are

visible and Environmental Line appears as if it were the root of the Products dimension.


118,99361,77840,95416,261Alert Devices

259,542123,44295,32740,773Bio-Friendly Soaps

56,96828,02019,3619,587Recycled Products

228,78495,61791,72041,447Sunblock

418,307206,054135,41076,843Water Purifiers

1,082,594514,911382,772184,911Environmental

Line

Individually, UC6 sees only Outdoor Products, whereas UC7 sees only Asia Pacific.

ProductsGO Sport LineEnvironmentalLine

OutdoorProducts

UC7

173,02719,12099,22254,685Australia

62,41924,90420,88716,628Hong Kong

70,60514,35227,03729,216Japan

75,00213,90437,76523,333Singapore

381,05372,280184,911123,862Asia Pacific

If multiple restrictions are applied to a dimension, the following occurs:

● UC5+UC6: members of both, logged on as <All User Classes>, see both Outdoor Products and

Environmental Line, because each have access to this data within one or more individual user

classes.






1,082,594514,911382,772184,911Environmental

Line

1,963,047933,988720,286308,773Products

● UC5+UC7: members of both, logged on as <All User Classes>, cannot see Outdoor Products

in Europe or North America, nor GO Sport Line in Europe or North America, because this

data is not accessible by either of these individual user classes.


OutdoorProducts

UC5+UC7

381,05372,280184,911123,862Asia Pacific

514,911denied382,772deniedEurope

514,911denied514,911deniedNorth America

1,278,73672,2801,082,594123,862Locations

● UC6+UC8: members of both, logged on as <All User Classes>, cannot see GO Sport Line, nor

Environmental Line in Europe or North America, because this data is not accessible by either

of these individual user classes.

ProductsOutdoor ProductsEnvironmental LineUC6+UC8

308,773123,862184,911Asia Pacific

337,514337,514deniedEurope

419,077419,077deniedNorth America

1,065,364880,453184,911Locations

Scenario Involving OmitThis scenario involves the following user classes and applied views:

● UC9 - Omit Products

● UC10 - Omit Locations

Individually, UC9 does not know that the Products dimension exists, because it does not appear in

the Dimension viewer. The user sees the Locations dimension and summary values across it.



Years20052004UC9

381,053190,6082190,445Asia Pacific

813,550448,716364,834Europe

1,070,660638,971431,689North America

2,265,2631,278,295986,968Locations

Individually, UC10 does not know that the Locations dimension exists, because it does not appear

in the Dimension viewer. The user sees the Products dimension and summary values across it.


OutdoorProducts

UC10

986,968137,976397,830451,1622004

1,278,295164,240684,764429,2912005

2,265,263302,2161,082,594880,453Years

In the union of user classes, UC9+UC10 log on as <All User Classes>. They do not see data for any

combination of Product and Locations. For example, neither can see how much revenue Outdoor

Products generated in the Asia Pacific. However, UC9 can see that the Asia Pacific generated revenue

of 381,053 in all years. UC10 can see that Outdoor Products generated revenue of 880,453. The

summaries are visible for the union of these user classes, as follows.


OutdoorProducts

UC9+UC10

381,053denieddenieddeniedAsia Pacific

813,550denieddenieddeniedEurope

1,070,660denieddenieddeniedNorth America

2,265,263302,2161,082,594880,453Locations

Scenarios Involving Parent-Child RelationshipsFurther complexities arise with the union of user classes when parent, child, and sibling combinations

are involved, as discussed below.



Parent-Child Category Relationships

The following table shows what an "all user classes" user sees in a dimension where various actions

are applied to categories in a parent-child hierarchy. This user belongs to two user classes. In one

arbitrary user class, actions are applied to the parent category. In the other user, actions are applied

to the child category. If it is not specified in the table, child values are rolled-up.

Parent Category

ApexSummarizeCloakExcludeChild Cat-

egory

No restrictions.The child is hid-

den.

The child is hid-

den. The parent

rollup values don't

include that of this

child.

The child is hidden.

The parent rollup

values don't include

that of this child.

Exclude

No restrictions.The child is hid-

den.

The child is hid-

den.

The child is hidden.Cloak

No restrictions.The child is vis-

ible but its chil-

dren are hidden.

The child is visible

but its children are

hidden.

The child is visible

but its children are

hidden.

Summarize

The apex category

is the nearest com-

mon ancestor. (See

"Union of Apexes"

below.)

The child is vis-

ible but its sib-

lings are hidden.

The parent is hid-

den. The child is

moved to be a

child of the

nearest common

ancestor.

The parent is hid-

den. The child is

moved to be a child

of the nearest com-

mon ancestor. The

common ancestor

rollup values don't

include those of the

siblings of the child.

Apex

Exclude

If a child category is excluded in one user class and its parent is hidden, that is, excluded or cloaked,

in another user class, the resulting view does not roll up the excluded child values into the parent,

as would normally be the case for a cloaked parent. This prevents unauthorized users from computing

the value of the excluded child by subtracting the sum of the parent's other visible children from

the parent's value. Note that due to this exception of suppressing the rollup, it is possible that the

value of a rolled up parent category may differ depending on the user class.

If a parent category is summarized in one user class and its child's children categories are excluded

in another user class, then the child's children's data is excluded in the union of user classes.



Union of Apexes

When user classes are apexed on different categories, the union of the user classes apex becomes

the nearest ancestor common to all the user classes. For example, consider the following dimension

hierarchy:

Locations

World

North America

Canada

Pacific Region

Central Region

Eastern Region

Atlantic Region

Northern Region

USA

Mexico

Europe

Asia

One user class is apexed on USA, and another user class is apexed on Eastern Region. The nearest

common ancestor is North America, which becomes the visible root category of the Locations

dimension. In addition, this new apex now has only two visible children, which are the apexes of

the individual user classes: USA and Eastern Region. Therefore, values from outside the new hier-

archy, such as Mexico, are not rolled up into the calculation for North America.

Scenarios Involving Multidimensional RelationshipsWhen restrictions apply to different dimensions, there is a smaller chance of having any restrictions

at all for cubes accessed using the <All User Classes> option. This is because as the number of user

classes increase, it is more likely that union of the user classes is allowed to see all the categories.

However, in cases were restrictions intersect, some data may be hidden. For example, consider the

Locations dimension above and the following Date dimension:

Date

Year

Quarter Fiscal Quarter (alternate drill down)

Month

One user class applies an exclude action to North America, and another user class applied an exclude

action to Fiscal Quarter. The union of user classes is not able to see a data value for the cell inter-

secting North America with Fiscal Quarter (plus any other intersecting category), since neither

individual user can see that value.



If you applied an apex on North America for one user class and applied an apex on Fiscal Quarter

1 for another user class, then the resulting union of user classes is able to see data from all fiscal

quarters for North America and data for all continents in Fiscal Quarter 1.

Alternate Drill-Down Paths in Time DimensionsWhen you create alternate drill-down paths in a time dimension, you must define the drill-down

paths so that Transformer can connect them appropriately at the convergence level. In particular,

the categories that descend from the convergence level must be identical. If you construct a time

dimension that contains multiple drill-down paths and you try to connect it at a level in which

descendant time periods do not exactly coincide, Transformer will be unable to connect the paths

correctly.

Several factors affect the way in which Transformer creates categories in levels containing dates,

and therefore potentially limit the ability to connect alternate drill-down paths in a time dimension.

The factors are:

● whether the drill-down paths differ in type (calendar or lunar time)

● how the Year Begins property is set for each drill-down path

● how the properties that affect weeks are defined in each drill-down path, such as the Week

Begins property, the Add an Extra Week property (lunar time dimensions only), and the Partial

Weeks property (calendar time dimensions only)

You must observe the following rules when constructing alternate drill-down paths in the time

dimension:

● You can create a time dimension with drill-down paths for both calendar and lunar time periods

only if the paths converge at the Day level.

● The Year level cannot be a convergence level for alternate drill-down paths.

● In all alternate drill-down paths, the Year Begins property must be set so that the convergence

level categories (Months, for example) coincide in each drill-down path. For example, to connect

alternate drill-down paths at the Quarter level, the Year Begins property in each alternate drill-

down path must be offset by exactly 'n' Quarters, where n is 1, 2, or 3 (the offset is less than

a full year). Similarly, to connect at the Month level, the Year Begins property in each alternate

drill-down path must be offset by an integer greater than or equal to one month and less than

or equal to 12 months.

You can visualize the concept by lining up the categories at the convergence level as in the fol-

lowing diagram.

The calendar year begins on January 1, 2000. The fiscal year begins on April 1, 2000, which

is one offset, one quarter from the start of the calendar year. The offset allows Quarter to be

the convergence level. When the Year Begins date is offset by integral units of the convergence

level, Transformer can construct descendant categories of the convergence level correctly.



● For both lunar and calendar time dimensions, the rules governing weeks must coincide in all

drill-down paths. In calendar drill-down paths, the Partial Weeks property must be identical

in each drill-down path. Similarly, in lunar time dimensions, the Add an Extra Week property

must be identical in each drill-down path.

Note: When you create lunar and calendar alternate drill-down paths, Transformer adopts the

labeling of the primary drill-down path for both time dimensions, and displays a warning message

that the labeling of the alternate drill-down path will conform to that of the primary drill-down

path.

Dimension Views in Alternate Drill-Down PathsDepending on your PowerPlay edition, portions of this topic may not apply to you. More more


and Editions topic in the Welcome chapter.

Because alternate drill-down paths share category values at and below the convergence level, you

must be careful when you create dimension views in one of the paths of a dimension that contains

alternate drill-down paths.

Any diagram command (Apex, Exclude, Cloak, Suppress, or Summarize) that you apply to categories

in one path will affect all other paths in the drill-down structure. Listed below are the effects of

performing diagram operations when you create dimension views in a dimension that contains

alternate drill-down paths:

Apex Operation

Eliminates other paths completely, if the apex category is either a drill category or below one of

the drill categories in the dimension. You can only reverse the effect of an Apex action by performing

another Apex action on the root category or drill category.

Exclude Operation

Excludes data from all other drill-down paths.

Cloak Operation

Removes categories from other drill-down paths, but allows their values to roll up to higher levels

in those paths.

Suppress Operation

For categories at and below the convergence level only, suppresses the shared category in all other

drill-down paths.

Summarize Operation

Removes child categories from other drill-down paths, but adds (or rolls up) their values to the

higher levels in those paths.




Example

In a Channels dimension, there are two alternate drill-down paths: By Customer Type and By

Region. If you create a dimension view that summarizes several Customer Types in order to focus

on Mass Marketers, Transformer eliminates the same categories in the By Region drill-down path.

When drilling down By Region in PowerPlay, the only Customer No categories available are those

that are Mass Marketers.

Current PeriodTransformer uses the current period setting to determine which date categories make up the relative

time categories in the time dimension.

For example, if a time dimension contains a relative time category called Last Month, and the current

period is set to April 1999, Transformer takes Last Month to be March 1999. If the current period

for the dimension changes, Transformer automatically updates the Last Month category.

Each time dimension can have its own current period, although all current periods will likely be

the same. The current period for a time dimension is shown on the Time tab of the Dimension

property sheet.

Process

You can set the current time period in four ways:

● automatically, based on the most recent date in the data source that updates the current period

● manually, by using the dimension diagram for the time dimension to choose a specific time

period as the current period

● on the command-line, by using the -t option

● in MDL, by executing the appropriate MDL script



How Data Sources Relate to the Dimension MapIn models created by using multiple data sources, each source is processed separately against the

model. Transformer doesn't perform database joins between data sources. Instead, Transformer

associates source columns directly with levels in the dimension map. If a level is unique, then

Transformer can match values from input records directly to the lowest levels in the dimension

hierarchy, without processing the entire dimension. If a level is not specified as unique, Transformer

must associate categories with the appropriate source values based on their relationship with other

columns in the same data source.

If a data source that contains the column for a level also contains all columns required for the

ancestor levels, you do not have to specify uniqueness. In this case, Transformer can identify the

categories in the level from the context afforded by the other columns in the data source. However,

when the columns for the ancestor levels come from different data sources, the values for categories

in that level must be specified as unique. Uniqueness ensures that Transformer can distinguish all

the categories in a level, without knowing their context.

For Transformer to relate columns from different data sources to the same level in a dimension,

the column names must be identical in all data sources.

Example of a Non-Unique Level Declared Unique

The following example illustrates how non-unique levels that you have declared unique on the Level

property sheet can cause problems in the model structure. The problems arise when the same category

value appears under two distinct parents, but the categories represent different entities.

In a Location dimension that contains the levels Country, State, and City, a source value based on

the city name does not necessarily ensure uniqueness. There might be more than one city with the

name Burlington (for example, one in each of the states Illinois, Massachusetts, and Vermont).

Additional context is required to determine which instance of Burlington is being referenced. If a

data record has all the fields required to identify the city (COUNTRY=USA, STATE=Illinois, and

CITY=Burlington), there is no ambiguity as to which Burlington is being referenced. Transformer

finds the right one by starting at the root category, finding USA, finding Illinois, and then finding

the one Burlington in that state.

Data source structure

Sample source data

COUNTRY,STATE,CITY

USA,Illinois,Burlington

USA,Massachussetts,Burlington

USA,Vermont,Burlington



However, if the data sources are set up so that values for the Country, State, and City levels come

from one data source while values for City, Outlet, and Sale Amount come from another data

source, the values for City must be unique. For example, assume that the values from the above

data source are derived from separate data sources as follows.


Sample source data

COUNTRY,STATE,CITY

USA,Illinois,Burlington

USA,Massachusetts,Burlington

USA,Vermont,Burlington

CITY,OUTLET,SALE_AMT

Burlington,Store01,10000



If the City level is declared unique, Transformer only creates a single "Burlington" category, and

associates measure values for each "Burlington" record found with that "Burlington" category, but

this leads to invalid results.

To avoid this problem in Transformer, you can use a calculated column to concatenate City with

State. Alternatively, you can change the source values for non-unique cities in the two data sources

by using qualifiers to make them unique:


Sample source data

COUNTRY,STATE,CITY

USA,Illinois,Burlington (IL)

USA,Massachusetts,Burlington (MA)

USA,Vermont,Burlington (VT)



CITY,OUTLET,SALE_AMT

Burlington (IL),Store01,10000

Burlington (MA),Store02,23000

Burlington (VT),Store05,25000

Example of a Non-Unique Category

The level uniqueness property does not necessarily mean that the real-world entities referenced in

the model are unique. In this example, a Customers dimension consists of the levels Country, State,

City, and Customer. The Customer level includes categories for individual customers in the cities

in the dimension:

Even though source data records may contain values for what is logically viewed as a single customer

(Acme Metals, for example), the level Customers cannot necessarily be considered unique. If that

customer (Acme Metals) transacts business in Chicago, Los Angeles, and Toronto, then there will

be three categories named Acme Metals in the Customer level but in different cities. This makes

the Customers level non-unique.

In order to uniquely identify each instance of Acme Metals, you must create a calculated column

that concatenates customer and city (in Transformer) or add qualifiers to the non-unique customers

(in your source data).

Tips

● To improve processing efficiency in models that use multiple data sources, ensure that all cat-

egories are uniquely defined in the source. Do not expect Transformer to relate values in one

data source with values in another by making joins between the columns.

● To accurately associate measure values from two different data sources, base your model on a

source column that generates categories for at least one level in the model.

Uniqueness Issues When Drilling Cube to CubeBackground

In Transformer, model dimension names and category codes must be unique. (For information

about uniqueness, see "Level Uniqueness" (p. 212).) Ordinarily, category codes default to the name

by which a category is referenced in the data source (the name by which the real world entity is

known in the database). This name must be unique in the context of its ancestor so that a data

record can be connected unambiguously to the category.

If there is more than one category in the same level or in different levels of the same dimension,

Transformer makes the category code unique by appending a tilde (~) and a number to the non-



unique source value. When you drill cube to cube, the context is carried from the original cube to

the destination cube by matching category codes in the corresponding dimensions.

Where category codes have unique source values, or the cubes are created from the same model,

there is no problem in matching categories. However, where cubes are created from different

models, and categories have codes made unique by appended numbers, categories that are actually

the same will not be matched because their appended numbers differ. In other cases, the wrong

categories may be matched.

When implementing cube-to-cube drill-through in cubes built from different models, especially

involving cubes created in third-party OLAP sources such as Essbase, ensure that category codes

in every dimension of every cube are unique and predictable. If your source values do not provide

unique category codes, create them by using calculated columns and specify uniqueness in the

Category Code Column of the Level property sheet.

Level PropertiesMultiple source values are associated with a level. Using Level Properties you can set the source

values of the levels in your model.

Source

Identifies the data source for the level. Levels in Transformer can be populated by multiple data

sources to improve performance in models with high data volumes.

Label

Identifies the name of the data source.

Description

Provides information in the PowerPlay Explain window.

Short name

Identifies an abbreviation of a long name for a dimension or category.

Category code

Identifies a category within a dimension.

Drill Through

Identifies the drill-through path for the level.



Appendix B: Functions

HIDD_COLUMNGENERAL The Calculation dialog box contains a common expression editor

that is shared with other components in the IBM Cognos Business Intelligence product suite. You

select elements one by one from the set of available expressions in the left pane of the dialog, and

build your expression in the right pane. As you enter each element, the syntax is evaluated against

standard mathematical rules.

The available elements vary according to the type of calculation and the position in the expression,

and each is explained in a ToolTip window in the Calculation dialog box.

You can only use components that appear in the Available Components list. However, you may be

able to access additional Expression Editor functions, such as characters-to-integer or dateDayDiff,

by creating the calculations in another supported data source product, and then importing them

into Transformer.

For more information on the functions, operators, constants, and components provided by the

common expression editor, see the Expression Editor online help.

change(cat_code, cat_code|set |level)Returns the difference over a linear series of categories by calculating the change between each pair

of categories. The order of pairs is based on the object ID of each category.

Note: This function is used for automatically generated relative time categories, such as year-to-

date (YTD Change). It appears in the list of Available Components when you use the Expression

Editor to create a calculated category. It is the same as the variance Summary Function in the

Expression Editor help.

Example

A dimension contains a level with Quarters Q1 to Q4. To create change categories for Q1-Q2, Q2-

Q3, and Q3-Q4, use the function

change("Quarter")

Tip: To use ToolTip on the pop-up diagram to show the labels and object IDs of categories, select

the Object Identifier check box in the Preferences dialog box (Titles tab).

percent-growth(cat_code, cat_code|set |level)Returns the growth over a linear series of categories by calculating the percentage change between

each pair of categories. The order of pairs is based on the object ID of each category.

Note: This function appears in the list of Available Components when you use the Expression Editor

to create a calculated category.


Example

A dimension contains a level with Quarters Q1 to Q4. To create growth categories for Q1-Q2, Q2-

Q3, and Q3-Q4, use the function

growth(Quarter)

Tip: To use ToolTip on the pop-up diagram to show the labels and object IDs of categories, select

the Object Identifier check box in the Preferences dialog box (Titles tab).

share(cat_code, cat_code|set |cat_code, set|set|level)Returns the values of categories as a percent share of other target categories.

Example

A dimension contains the category Color with several descendant categories, such as Black and

Chrome, which you define as a set. To create two share categories, representing each of Black and

Chrome, as a percentage share of the parent, use the function

share(set, Color)

Note: This is the same as the percentage Summary Function in the Expression Editor help.

Parameterscat_code

The category code string specified in the property sheet for the category.

set

The category set defined in the Category Calculation dialog box. Define a category set by selecting

categories in the pop-up dimension diagram and dragging them to the Category Calculation dialog

box.

level

A level in the same dimension. Select a level from the pop-up menu and use it in your expression

editor calculation.


Appendix B: Functions

Appendix C: Command Line Options

You can use command line options to perform routine tasks, such as

● creating or updating cubes using the appropriate executable

● updating a model with new categories generated during the category generation process

● running a set of batch jobs with different preference settings or files

● changing the current date setting so that relative time calculations in batch cube creation are

based on a specific date

● supplying database signon information

● changing the degree of detail for log file messages

● opening and deleting checkpoint files

● verifying and, if necessary, updating the scales used in MDL model columns and measures to

match those used in the datasource

Windows-only command line options enable you to

● run Autodesign based on an .iqd (Impromptu Query Definition) data source, build a model,

create a cube, open PowerPlay, and display a report

● open the Transformer executable without showing a splash screen

● regenerate categories, and the measure scales used with them, without building the cube

UNIX-only command line options enable you to

● open specified .mdl files and/or accept MDL statements

● specify the number of records for a test cube

● prepare to accept input from the standard input stream

The syntax for using the Windows command line interface is as follows:

Trnsfrmr -<option><model_file> where <model_file> can be a .py? or .mdl file type.

The UNIX syntax is as follows:

rsserver <options><model_file> where you specify the processing information by means of a

dash or by one or more case-sensitive options.

Notes

● Although the UNIX command line syntax is case-sensitive, the Windows syntax is not.

● You can combine command line options. For example:

Trnsfrmr-n -i Sales.py?


Trnsfrmr -i -t2006 Sales.py?

● The space character is used to separate multiple command line options. You cannot add spaces

between an option and its argument, and if any of the arguments contain spaces, you must

enclose them in double quotes. For example:

Trnsfrmr -n -kfieldthree=CarlosR/pw462 Field3.mdl

is not valid, whereas

Trnsfrmr -n -k"field three=CarlosR/pw462" Field3.mdl

is valid.

● When you run a batch file to create a model in Transformer, the code returned is zero if suc-

cessful and non-zero if unsuccessful. A non-zero error code may be returned in a number of

circumstances, such as failure in the cube creation or update, or if the application terminates

unexpectedly. The following Windows batch file tests whether the operation ended successfully:

cd <install_dir>:\ProgramFiles\Cognos\cern\bin

trnsfrmr.exe -n2 -s -i -nologo <install_dir>:\<filename>.py?

if errorlevel 1 goto error

:noerror

echo no error was encountered

pause

goto end

:error

echo Error

pause

:end

● On UNIX, you can append the following Bourne Shell script to your rsserver command, to

perform another action on successful completion:

#!/bin/sh

if rsserver<command line options>

then

#perform action a if exit status is 0

else

#perform action b

fi

Command Line Options ListThe available Windows and UNIX command line options are summarized below, and explained

in detail in the sections that follow:



MeaningRestrictionsOption

runs AutoDesign for an IQD-based

model, then creates a cube, opens

PowerPlay for Windows and dis-

plays a report

Windows only-a <data_source>

generates categories and creates

cubes using rsserver

UNIX only; used with -i, -m,

or -p

-c

checks for and logs obsolete user

classes found in an existing model

Removal of objects and views

is done on Windows UI

-cs .mdl_file

overrides user preference settingOn UNIX, demands an upper-

case D; no space between the

option and its argument

-D<preference_var>=<set-

ting> .py?_file |.mdl_file

updates model but not dataCannot be used with -c;

should always be used with -

n and -o

-e .py?_file |.mdl_file

sets user-defined preference fileOn UNIX, demands an upper-

case F; no space between the


-F<preference_file> .py?_file

|.mdl_file

opens specified .py? model and

restarts failed process from begin-

ning

On UNIX, cannot be used

with -s

-i .py?_file

specifies database signonNo space between the option

and its argument; depending

on your edition of Trans-

former, may not apply. For

more information, see the

Features and Editions topic.

-k<signon>=<userid>/

<password> .py?_file

|.mdl_file

opens specified .mdl file or accepts

Model Definition Language (MDL)

statements

UNIX only-m .mdl_file

runs Trnsfrmr.exe in batch modeWindows only-n<display_state> .py?_file

|.mdl_file

omits splash screen display on

opening

Windows only-nologo .py?_file |.mdl_file



MeaningRestrictionsOption

turns off various model and cube

creation actions; see description

-ox is Windows only-o .py?_file |.mdl_file and -

ox

opens specified .py? fileNot valid with an MDL file

and should not be used with -

s

-p .py?_file

specifies log level detailValid levels are 0, 1, 2, 3, and

4; no space between the


-r<log_level> .py?_file

|.mdl_file

saves modelOn Windows, used in conjunc-

tion with -n; cannot be used

with -i or -p

-s .py?_file |.mdl_file

sets current periodOn Windows, used in conjunc-

tion with -n; no space between

the option and its argument

-t<category_code> .py?_file

|.mdl_file

gets partition statusCubes must have been gener-

ated; does not work on

secured cubes; no space

between the option and its

argument

-u<powercube_name>

.py?_file |.mdl_file

specifies number of records for test

cube

UNIX only; used with -c, -m,

or -p

-v data_subset_number

updates column and measure scales

based on the data source

Slightly different behavior on

UNIX than on Windows

-x .mdl_file

expect input from the standard

input stream

UNIX only- (dash)

-a <data_source>A Series 7 Transformer for Windows option. Runs AutoDesign, based on the specified data source,

to build a model, then create a cube, open PowerPlay, and display a report.

Example: WindowsTrnsfrmr -a <Sales_datasource>

Opens the Sales data source and places dates in the time dimension, columns with numerical values

in the Measures list, and all remaining columns in the dimension map, by using the best-fit approach.

Then, creates the Sales cube, opens PowerPlay, and displays the Sales report.



Notes

● You can only use this option with the -d, -f, -r, and -nologo options.

● The -k option, which you can normally use to supply Transformer with a signon in batch mode,

should not be used with IQD datasources or with autodesign. In these cases, a logon screen

appears, and a user must intervene to supply the required signon information, such as the logical

name that appears in the IQD source.

● There must be a space between -a and the datasource file name.

-cA UNIX-only option. Generates categories and creates cubes after rsserver loads a model file,

interprets MDL statements, or both.

You must use this option with the -p, -m, or -i option.

Example: Generate Categories and Create Cubes on UNIX

In both of the following examples, the command starts rsserver and builds the cubes that the model

specifies. The .mdl file is a full model definition.

● This command processes a binary model file (.py?) called go_sales.py?:

rsserver -c -p go_sales.py?

● This command processes an MDL text file called go_sales.mdl:

rsserver-c -m go_sales.mdl

-cs .mdl_fileChecks for and logs obsolete user classes found in an existing model. The namespaces checked are

those accessible using the specified userid and password.

Example: WindowsTrnsfrmr -m -cs go_sales.mdl -user <userid> -pw <password>

Logs on as an authenticated user and, for each namespace associated with the go_sales model that

is accessible to the specified <userid> and <password>, outputs a log file listing the obsolete user

classes.

Example: UNIXrsserver -m -cs go_sales.mdl -user <userid> -pw <password>

On provision of the appropriate user ID and password, checks for and lists the obsolete user classes.

Notes

● Used with -m in order to specify that the ASCII form of the model file (.mdl) is to be parsed.

● To remove obsolete objects and their associated views, you must log on to the Windows UI.



-D<preference_var>=<setting> .py?_file |.mdl_fileSets a new value for a Transformer preference.

The value overrides settings from the Preferences dialog box for this instance only. Also, if you

specify the -D option after the -F option, the -D setting overrides the setting in the -F file, and vice

versa.

On UNIX, an uppercase D is required. No space may appear between the -D and the argument

<preference_var>. All <preference_var> settings are case-sensitive and must match the values specified

in the trnsfrmr.ini file. For example, you must use -DLogFileName not dlogfilename.

Example: Override Preference File Settings on WindowsTrnsfrmr -dDataSourceDirectory=C:\Newdata -n Transact.mdl

Sets the Data Source Directory to be C:\Newdata, opens the file Transact.mdl, then runs the process

in batch to create all defined cubes.

Example: Override Preference File Settings on UNIXrsserver -DChildRatioThreshold=25

Overrides the default value at which rsserver issues warnings that a category has too many descend-

ants.

Notes

● In Series 7 Transformer, you can use most of the settings in your Trnsfrmr.ini file as arguments

for <preference_var>. Trnsfrmr.ini is located in the executable directory of your most recently

installed rendition of the product. The setup program moves any existing Trnsfrmr.ini file to

this spot, from the Windows directory or the home directory of a multi-user Windows system,

to retain your settings from previous versions of Transformer. Any settings in the [PowerPlay

Transformer] section of Cognos.ini are read; however, settings in Trnsfrmr.ini override those

in Cognos.ini.

● You can include several -d options on the command line. However, for ongoing production

use, you can instead create your own preference file, based on Trnsfrmr.ini, and have Trans-

former access your file with the -f command line option. You can also use environment variables

to set preferences. See "-F<preference_file> .py?_file |.mdl_file" (p. 247).

-e .py?_file |.mdl_fileUpdates and saves all the cube metadata that is defined in the model, but does not update the data.

The cube metadata consists of object names, labels, short names, descriptions, drill through reports,

and user classes.

You cannot use this option with -c, and you should always use it in combination with the -n and

-o options, to run in batch and suppresses cube creation. Otherwise, you defeat the purpose of this

command.

Example: WindowsTrnsfrmr -e -n -o New.py?



Opens the model New.py?, updates the defined cubes without regenerating existing categories or

creating new ones, including all of the cube metadata (object names, labels, descriptions, drill-

through reports, user classes), and then saves the model file.

-F<preference_file> .py?_file |.mdl_fileSpecifies the user-defined preference file or files to use. The name may include a directory path. No

space may appear between -F and the argument <preference_file>. On UNIX, an upper case F is

required.

If you are performing batch tasks and need to specify multiple preference files, Transformer combines

the settings in each file successively; later settings override previously defined ones.

Example: Specify a Preference File on WindowsTrnsfrmr -n -fc:\Monthly.prf Transact.mdl

Sets the preference file to C:\Monthly.prf, opens the file Transact.mdl, then runs the process in

batch to create all defined cubes.

Example: Specify a Preference File on UNIXrsserver -Fmypref.prf -m go_sales.mdl

Commands rsserver to parse an .mdl file using the preference file mypref.prf.

Note

If the preference filename does not include a path, Series 7 Transformer searches the executable

directory of your most recently installed rendition of the product. The setup program moves any

existing Trnsfrmr.ini file to this spot, from the Windows directory or the home directory of a multi-

user Windows system, to retain your settings from previous versions of Transformer. Any settings

in the [PowerPlay Transformer] section of Cognos.ini are read; however, settings in Trnsfrmr.ini

override those in Cognos.ini. See also "-D<preference_var>=<setting> .py?_file |.mdl_file" (p. 246).

-i .py?_fileOpens a saved model, regardless of the existence of a checkpoint file.

A checkpoint file exists if a model is suspended due to a general system failure, such as that caused

by a power outage. Normally, you are offered the choice of opening the checkpoint file or the last-

saved version of the model file, the next time the model is opened. However, you can use the -i

option when working in batch mode to bypass the prompt and force Transformer to open the original

model file instead of the checkpoint file.

In UNIX, cannot be used with the -s option.

Example: WindowsTrnsfrmr -n -i Sales.py?

Opens the model file Sales.py?, discarding any existing checkpoint file, then runs the process in

batch to create all defined cubes.



Notes

● Checkpoint files have a .qy? extension. The ? (question mark) in the extension .qy? is replaced

by the character that is used in your release of Transformer.

● For more information about error recovery, see "Recover a Failed Model" (p. 158).

-k<signon>=<userid>/<password> .py?_file |.mdl_fileIn Series 7 Transformer, supplies the signon information needed to establish one or more database

connections during batch processing. It provides an alternative to storing the information in the

Signons list or using Access Manager to retrieve user IDs and passwords. This is especially useful

when used with .mdl files which, by default, don't use "verb MDL" and so do not store signon

passwords.

For databases referenced in an Impromptu query, the signon is the logical database name that

appears in the .iqd file and the Cognos.ini file. Multiple .iqd data sources can use the same signon

object. You can view these signons on the Series 7 Transformer Windows interface, but you cannot

change them.

Example: WindowsTrnsfrmr -n -kfield=CarlosR/pw462 Field3.mdl

Opens the file Field3.mdl and grants access to the required database using the signon named field,

user ID CarlosR, and password pw462. If the signon matches that defined in the model, the process

runs in batch mode, creating the cubes as defined in the model.

Notes

● No space may appear between -k and the argument <signon>.

● The signon name cannot be empty or contain the ASCII equal (=) character.

● The user ID cannot contain the slash (/) character.

Example: UNIX

In this example, Transformer reads your data source for model Xyzsales.mdl from an Oracle

database using an .iqd file. The signon sal_log includes the Oracle user ID corpadm and the password

my_pass. The command to process the .mdl file for model Xyzsales.mdl is as follows:

rsserver -c -s -m Xyzsales.mdl -ksal_log=corpadm/my_pass

If you use the -k option to pass user IDs and passwords to rsserver, there is a possibility that

security breaches may occur. For example, your security can be violated if other users can view the

details of your rsserver process using the ps command or by reading the log file.

To prevent security problems, you have these options:

1. Store the -k option with the user ID and password information in a file in a secure directory

and call the file from the rsserver command line.

2. Embed the user ID (or the user ID and password) in the MDL model file.

3. Create an MDL script file.



In the first case, you can create a file called Sal_id.txt and store it in a secure directory. This file

contains the -k option with user IDs corpadm and corpis and passwords my_pass and bld_cube.

-ksal_log=corpadm/my_pass -ksal_cube=corpis/bld_cube

You can then call the Sal_id.txt file from rsserver as follows:

rsserver -c -s -m Xyzsales.mdl 'cat Sal_id.txt'

In the second case, you can add MDL statements to the end of the .mdl model file. These embed

the user ID and password, and update the signon information needed to log on to the database:

SignonUpdate "sal_cube" PromptForPassword False UserID"corpis" Password "bld_cube"

To embed the user ID only, enter:

SignonUpdate "sal_cube" PromptForPassword True UserID"corpis"

You can then run rsserver with the -m option, specifying the modified .mdl file without the -k

option.

In the third case, you can create a secured temporary MDL script on the server. The MDL script

updates the model signon, as shown in this example:

OpenPY "Xyzsales.py?"

SignonUpdate "sal_cube" PromptForPassword False Password"bld_cube"

SavePY "Xyzsales.py?"

To use this MDL script, run rsserver with the -m option, specifying the name of the .mdl file.

-m .mdl_fileOn UNIX only, specifies an ASCII model or script file (.mdl) to interpret.

If you use multiple occurrences of -m, files are processed in the order of their occurrence. In addition,

you can use a dash (-) as an argument to accept MDL verb statements from the standard input

stream.

Example: Save a File in MDL Format on UNIX

1. Create a separate file, Savemdl.mdl, containing the line:

SaveMDL"Xyznew.mdl"

2. Using the ASCII model file Xyzsales.mdl, type the following command:

rsserver -m Xyzsales.mdl -m Savemdl.mdl

Then, using the binary model file Xyzsales.py?, type the following command:

rsserver -p Xyzsales.py? -m Savemdl.mdl

3. Input SaveMDL from the standard input stream by typing the following commands:

rsserver -p Xyzsales.py? -m

SaveMDL "Savemdl.mdl"

4. Finish with the UNIX end-of-file command (Ctrl-D) to save the .py? file in .mdl format.



-n<display_state> .py?_file |.mdl_fileRuns Transformer in batch mode. This option is not valid for UNIX.

Use this option to open Transformer, create cubes identified in the model, and close Transformer.

You can optionally specify a window display state of 1 to minimize the Transformer window, or

2 to hide the Transformer window.

Example: WindowsTrnsfrmr -n -s Roofing.mdl

Opens the model file Roofing.mdl, creates its defined cubes, then saves the model file.

Trnsfrmr -n2 Roofing.mdl

Opens the model file Roofing.mdl without displaying an application window, then creates its defined

cubes.

-nologo .py?_file |.mdl_filePrevents the display of the Transformer splash screen.

Example: WindowsTrnsfrmr -nologo Roofing.mdl

Opens the model Roofing.mdl without first showing the splash screen.

-o .py?_file |.mdl_fileOn Windows, regenerates the categories after a model load, but disables cube creation. For batch

mode operations, can be used with the -n option to open Transformer (Windows), perform the

specified action, and then close Transformer.

A separate Windows-only option, -ox (which is not a combination of -o with -x) disables both

category creation and the cube rebuild.

On UNIX, turns off both population of the model and cube creation, after the .mdl file is loaded.

Example: WindowsTrnsfrmr -s -n -o Roofing.mdl

Opens the model file Roofing.mdl, and saves any model changes, such as new categories. Transformer

then closes without creating any cubes defined in the model.

Example: UNIX

rsserver -o go_sales.mdl

After the file go_sales.mdl is loaded, disables both population of the model and cube creation.

-p .py?_fileOpens the checkpoint file (.qy? on Windows) or loads a binary model file (.py? on UNIX) and

processes it, beginning from the last checkpoint saved in the checkpoint file if this exists, or beginning

from the start of the .py? file. All changes are saved on termination.



A checkpoint file exists if model processing was suspended. It has a .qy? extension, where the ?

(question mark) is replaced by the character that is used in your release of Transformer.

Example: WindowsTrnsfrmr -n -p Sales.py?

Opens the checkpoint file associated with model file Sales.py?, then continues to process the model

in batch mode, creating all defined cubes.

Example: UNIXrsserver -p go_sales_jan.py? -m monthly_update.mdl-Ftrnsfrm_prd.prf

Starts rsserver, processes the commands (MDL verbs) in the file monthly_update.mdl, obtains

preferences from the file trnsfrm_prd.prf, and saves the model.

-r<log_level> .py?_file |.mdl_fileSets a degree of detail for messages written to the log file, where log_level is a value from 0 to 4.

Each level includes the errors and messages for the higher levels:

● 0 - the Enable Message Logging box is cleared; logging is suppressed

● 1 - includes only severe errors

● 2 - includes error messages and above

● 3 - includes warning messages and above

● 4 - includes informational messages and above (the default setting)

Example: WindowsTrnsfrmr -n -r2 -dLogFileName=Roof.log -dLogFileDirectory=F:\Test Roofing.mdl

Opens the model file Roofing.mdl, sets the degree of detail for messages to 2, then runs the process

in batch to create all defined cubes. The messages are written to the file F:\Test\Roof.log.

Notes

● No space may appear between -r and the argument <log_level>.

● For more information on error logging, see the reference topic dealing with the Log file.

-s .py?_file |.mdl_fileAfter successful creation of a cube, before closing Transformer, saves the model in binary format

(.py?), with any new categories added during the category generation process.

Example: WindowsTrnsfrmr -n -s Roofing.mdl

Opens the model file Roofing.mdl, creates its defined cubes, saves the model file, and closes

Transformer.



Example: UNIXrsserver -m go_sales.mdl -s go_sales.py?

Starts rsserver, parses an ASCII model file (.mdl), and saves the changes in a binary model file (.py?).

Notes

● In Windows, this option must be used in conjunction with -n.

● In Windows or UNIX, do not use this option with -i or -p.

-t<category_code> .py?_file |.mdl_fileSets the current period for the purpose of calibrating relative time calculations.

In Windows, this category code is defined on the Category property sheet. The category must exist

in a manually-set time dimension, which is set by clearing the Automatically Set Current Time

Period box in the Dimension property sheet (Time tab).

Example: WindowsTrnsfrmr -n -t"20051201-20051231" -s-Year3.mdl

Opens the model Year3.mdl, sets the current period to the category which has a category code of

20051201-20051231, then runs the process in batch to create all defined cubes.

Notes

● No space may appear between -t and the argument <category_code>. If the category contains

hyphens or space characters (as in the date range above), it must be enclosed in double quotes.

● The <category_code> portion of the command is case-sensitive; it must exactly match the cat-

egory code in the model.

● In Windows, this option must be used in conjunction with -n.

-u<powercube_name> .py?_file |.mdl_fileWrites the partition information for a specified cube to the log file.

The partition information is written in the following format:

<date> <time> Cube <PowerCube_name> partitionreport Partition #

Category Code Category Name Partition Size

Example: WindowsTrnsfrmr -i -n -uNorth -uEast -dLogFileName=F:Roof.logRoofing.mdl

Opens the model file Roofing.mdl, discarding any existing checkpoint file, creates all defined cubes,

and writes the partition information of cubes North and East to the log file F:\Roof.log.



Notes

● Cubes must be generated before their partition status can be reported. If the command line

includes options to generate categories and create cubes, those options are processed before

the partition information is obtained.

● To obtain the partition information of cubes in a cube group, you must specify the individual

cube names, not the cube group name.

● The cube name is case-sensitive. In the above example, you must type uNorth, not unorth, for

this command to work; no space between the option and its argument.

● Partition information cannot be provided for secured cubes and cube groups.

-v data_subset_numberSpecifies how many data source records rsserver should use to create a test cube. If you have a large

data source file, this option enables you to do a test run on a limited number of records before

processing the entire file.

If the number of records you specify is greater than the total number of records in the file, rsserver

treats the process as a normal run, not a test, and uses the whole file.

Use this option on UNIX only, in conjunction with the -c, -m, or -p option.

Example: Create a Test Cube from a Subset of Records on UNIXrsserver -c -p Xyzsales.py? -v 525

Commands rsserver to process 525 records from binary model file Xyzsales.py?, generate categories,

and create a cube.

-x .mdl_fileUpdates the column and measure scales of the MDL model, provided the data source can handle

queries about scale. This means that the option is supported for relational or ADO data sources,

including Microsoft Access and Microsoft Excel, but not ASCII or other flat-file data sources. Begins

by verifying that all column scales match those in the source. Then, checks all of the associated

measures, updating their output scales as required.

- (dash)On UNIX, use to have rsserver accept input from the standard input stream. Any options listed on

the right of the dash are ignored.



Appendix D: Proven Practices inModel and SystemDesign

This chapter describes proven practices you can adopt when modeling your PowerCubes and

designing your production system.

Recommendation - Analyze Your Requirements and Source DataThe first step in developing an effective business intelligence model is to analyze your users and the

reports they require. Use questions such as the following to help you analyze your users.

● What reports do users currently use? Which do they use most frequently? Which do they use

only rarely?

● Does each group require different reports? Are there some reports that are required by all users?

● Do users need detailed drill-through reports, summary reports, or both?

● How frequently are the measures in the report updated? How frequently do the reports them-

selves change? Does the frequency vary from group to group?

● How often are reports required? Can you trade off timeliness in order to ensure accuracy? For

example, if your users ask for monthly reports and the data source is refreshed weekly, the data

will always be current. However, if your users want daily reports, the data will only be up-to-

date on the first day of the weekly cycle.

Identify Your Security NeedsAssess the need for controlling access to information and, if required, identify the security levels of

each user class or group. Use questions such as the following to help you analyze security needs.

● What business information can be made available to all employees? Which information must

be restricted to certain individuals or groups?

● Can you place your users into distinct classes based on their information needs and access

privileges? Or do people move about so frequently that this is not practical?

● If your organization already has user groups, are these based on network hook-ups and database

access, or existing Human Resources classifications (job functions and task profiles)? Must you

realign these user groups to more accurately reflect decision-making roles?

● Can you rely on network operating system logins to control access, or must you implement

alternative security for your sensitive data? Do you have user-class or directory-based security

already in place?


Assess Your Source DataUse questions such as the following to help you analyze your source data.

● Does the data come from one source or many? What format is it in: flat files, Impromptu

reports, spreadsheets, or databases? Does it need to be converted to a supported source type

before it is imported?

● How many records are there? How much do you expect the data to increase?

● How much of the data is static and suitable for incremental updates? Can you create different

data sources for static and non-static data to support incremental updating? This shortens cube

creation time by appending new data to a cube instead of recreating it.

● How much data preparation and cleaning are required? (You should clean and preprocess your

data as much as possible.)

● If you are using a database, do you plan to use pre-summarized data?

● Are all data items unique?

● Are linked measures from different sources updated at the same time?

● Must you create additional data sources to accurately model your organization?

Before you begin designing your model, take a closer look at your data. Consider the relative

advantages of merging data into an Impromptu .iqd file, keeping it in separate files, or restructuring

it for better performance and flexibility.

Merge Data into an Impromptu Source File

Transformer can obtain input in a variety of formats, including flat files, spreadsheets, Access, and

Paradox. If the data you need is stored in a relational database that Transformer doesn't read,

depending on your edition of Transformer, you may be able to

● preprocess the data using a tool that creates comma-separated values (CSV) files or fixed-field

files

● create an Impromptu query definition file (.iqd) and then access the database directly or by

means of an ODBC connection. To ensure that Transformer connects to the database defined

in Impromptu in read-only mode, override the default isolation level for the database by

changing the IsolationLevel property of the Signon for your database in the Signons list in

Transformer.

● import CSV or Query Definition files exported from Framework Manager, the IBM Cognos 8

modeling component

Leave Data in Different Source Files

If several different source files provide the input for one Transformer model, you can customize the

resulting cubes to suit the needs of different users, and store the files in various secure locations. In

such models you can do one of the following:


Appendix D: Proven Practices in Model and System Design

● specify a different source file for each data source in your model. You change the source file

name, but retain the other properties associated with the data source. You can then specify

which data sources you want to use when you create each cube.

● use a separate data source to point to a source file whose contents change often. For example,

if your hospital admissions data changes weekly, put this information in a separate file, and

specify that this source be queried every time your cube is processed. That way, the cube will

always contain the most up-to-date information.

Presort, Clean, or Consolidate Your Data

Consider the performance benefits of preprocessing your data further:

● Presorted records are processed more quickly by Transformer.

● If you streamline your source data so that it only contains the information needed for the model,

read time is faster in Transformer.

● You can use Transformer to presummarize the data if your users don't need to access all the

details in the source.

● If your organization processes 50,000 transactions daily, for example, and you only create the

cube weekly, then summarize the transactions at the weekly level before Transformer begins

processing. This will greatly speed up cube creation.

● Consolidation (that is, combining records with identical non-measure values) reduces the size

of the cube and improves performance in your reporting application. Consolidation is enabled

by default in Transformer. Evaluate your data to see if it can be further consolidated by using

the duplicates or regular rollup features of Transformer. For consolidation purposes, non-

measure values are considered identical if they meet any of the following criteria for the partic-

ular rollup.

● The source data contains transactions with identical non-measure values. For example, two

sales of the same product are made to the same customer on the same day, but the colors differ.

If colors are omitted from a dimension view using the Suppress or Summarize command, the

sales records will have identical non-measure values.

● Records become identical when a dimension is omitted from the cube. For example, two sales

of the same product are made at different stores on the same day. If the Stores dimension is

removed from the model, these sales records will have identical non-measure values.

● Records become identical because of the Degree of Detail setting. For example, if the Degree

of Detail is set to Month for a column associated with a time dimension that includes week

and day values, Transformer ignores the week and day values in the source transactions when

consolidating records.

Separate Structural and Transactional Data

It is helpful to know whether your data sources contain purely structural information, transactional

information (measure values), or a combination. Processing time improves if you can query structural

and transactional information separately. Ideally, you should define each dimension or drill-down



path with a separate structural data source and add one or more transactional data sources to

provide measures for those dimensions.

We recommend defining the structural data sources in the same order they are used to build the

dimensions and levels, listed from left to right in the Data Source dialog box. This allows Transformer

to process one data source, update the associated dimension and its categories, and then continue

with the next data source. The last items on the list should be the transactional sources that provide

measure values for the dimensions.

If the data sources are not listed in the correct order, and optimization is not set to Default (auto-

partition), Transformer may have to process a data source repeatedly to build a single dimension.

We recommend defining any transactional data sources that change frequently so that they contain

a small, concise record set, with the minimum amount of information needed to update the data

in your PowerCubes. Whenever possible, maintain your category structures within your Transformer

model, to eliminate the redundant processing required to continually rebuild it.

Finally, if your models contain long descriptions, we recommend that you generate PowerCubes

using models that are already populated with categories associated with those descriptions.

The following subsections provide some additional tips and guidelines.

Analyze the Data Flow

Analyze the flow from the point at which your data is generated until the data is input into Trans-

former. Should the data be streamlined or rationalized at any point? If so, can you use a data

warehouse, a series of data marts, or a data-extract procedure to reorganize it?

Remove Any Constraints

You may have to resolve uniqueness issues and data dictionary terms before you can merge two

sets of data into one multiple-data source model. Ensure data integrity by checking your column

joins: outer joins or table aliases may be required. (Transformer isn't a relational database tool,

and doesn't perform joins between the columns of different data sources. If you need to set up

database joins, use Impromptu or QueryDefinition files exported from Framework Manager, the

modeling component of IBM Cognos 8.)

Build Flexibility into Your Plan

If you can, use a different source file for each aspect of your business, and organize the data sources

in your model so that each data source supplies the data for a different dimension. That way, you

can add more information into your cube as your business evolves, even if the data comes from

different software applications, platforms, departments, or locations.

Preprocess for Better Performance

Performance improves as you reduce processing loads in Transformer. One way to do this is to

partially normalize your source data by restructuring it into structural and transactional data sources.

This speeds up both category generation and cube creation.



Recommendation - Build a PrototypeThe best way to field-test the accuracy of your analysis is to build an initial model to reflect the

needs of key decision-makers in your enterprise. Base your pilot project on an existing set of popular,

stable reports and use the following as a set of signposts along the way.

Identify MeasuresMeasures are the numbers you use to gauge your organization's performance. You should choose

the critical success factors in your business to be your measures. Examples of typical measures

include sales revenues, profit margins, and response times.

If you have multiple data sources, the dimensions and levels of your model should be related to the

data source that contains the columns that will be used for your measures.

You will have a more effective model if your measures are applicable to more than one dimension.

For example, if your dimensions are products, locations, and customers, your measures should

bridge these dimensions.

Specify a Time DimensionTo ensure that your users can make period-to-period comparisons and visualize trends over time,

choose a time dimension that reflects and synchronizes accounting periods and reporting schedules.

In most cases, your users' needs are met by models based on the calendar or fiscal year. Month,

Quarter and Year categories are supplemented by automatically generated relative time categories

such as Percent Growth year-over-year.

However, many organizations need to use custom time periods, such as lunar weeks and months,

or 8-hour shifts. For more information about how to design custom time periods, see "Set Up a

Custom Time Dimension" (p. 85).

Select the Data to be Modeled (Sales Example)You need to set up your data sources so that they meet the needs of the Transformer model that

you want to create. For example, in your sales organization, information about customers is stored

in a Customers table, while information about products is stored in a Products table. Related tables,

Customer_Details and Product_Details, provide additional information about customers and

products. Order information is stored in two tables called Orders and Order_Details.

In keeping with good design practice, you decide to set up the Customers, Customer_Details,

Product, and Product_Details tables as structural data sources, to provide the information that

Transformer uses to build the Customers and Products dimensions in the model.

The information about transactions is stored in the Orders and Order_Details tables. For efficiency,

you decide to combine the information in these tables into a single data source called Order_Info.

The Order_Info data source contains all the information that Transformer requires to associate

sales with customers and products:

● the order date, to generate categories for the time dimension

● header information such as the customer, sales representative, and so on



● the product, order quantity, and sales amount for each line item in an order, to provide sales

measures

● the cost of the order and any discounts applied to it

You can create your initial dimension map on paper, to capture all of the dimensions, levels, cat-

egories, and measures needed in your cube.

CustomersLocationsProductsOrder Date

Market SegmentRegionProduct GroupYear

Customer TypeOfficeProduct ClassQuarter

Customer GroupSales RepProduct NameMonth

SubgroupProduct No.

Sales, Order Qty, Cost, DiscountSample Measures

Create and Refine Your ModelOnce you have created an initial dimension map on paper, you then create a new PowerCube model

in Transformer. We recommend that you begin with only one type of data source. Later, you can

enhance the business value of your model by adding more data sources, or by manipulating the

data derived from the existing data sources.

In the sales-tracking example, you might not initially have information about staffing levels at each

branch. If you add another data source or use the category count feature of Transformer to provide

these details, then a report based on your model can deliver value-added information on the average

sales per employee.

Models can contain any combination of

● regular measures: the numeric data found in a transaction file

● calculated measures: the numeric data calculated from other measures, mathematical operators,

and numeric constants

● category counts: the number of categories in a unique level for which the measure values are

non-missing or non-zero

● calculated categories: calculations that apply directly to any measure

● calculated columns: new data based on values calculated from other columns, functions, con-

stants, and calculated columns

If your expanded model incorporates multiple data sources, remember that Transformer must be

able to associate values from a data source column with categories in a level. Unless a data source

contains columns for a level and all of its ancestors, thereby establishing a context, the level must

be unique.



To refine your model, make some of these enhancements, if feasible:

❑ Add special categories to enable quicker data access.

Group your data into special categories based on attributes that may be contributing to the

success of your enterprise, such as product color or customer income.

❑ Add drill-down paths to provide more detail.

A dimension normally consists of a single drill-down path with one or more drill-down levels,

representing the hierarchical organization of the information. However, you can further subdivide

your dimensions, so your report users can analyze their data at different levels of detail.

There are no restrictions on the number of levels and drill-down paths that you use in a

dimension. However, all alternate drill-down paths must converge at a common unique level

and, for performance reasons, it is best to keep a 1:10 ratio or less between the categories in

each level.

❑ Allocate measures to other levels or dimensions.

If your model uses multiple data sources, consider allocating measures to levels or dimensions

with which they are not normally associated. Allocation can provide you with new insights

into your data. For example, you can associate resource-related data to financial data.

You can allocate measures over entire dimensions, over levels within an individual dimension,

or over categories within levels. When allocating, use measures that come directly from your

source data, rather than calculated measures, and take care not to overload your model with

superfluous detail.

❑ Incorporate data from another functional area.

Consider combining information from another functional area, such as materials and resource

planning or performance quality, with the finance or customer profitability data already in

your business model.

Begin by listing the data columns and by determining if there are any gaps, particularly in the

area of cost of materials, or indicators of quality.

Next, map the new dimensions, checking that the time periods are consistent with each other

and with your financial statements. Ensure that revenue and expense values map to those in

the financial statements.

Finally, verify that one-to-one relationships exist between the various measures or, if not, confirm

how each relates to your common dimensions.

Example - Add Customer Service Data to Refine your Model

Suppose your initial model includes the following:

TurnoverInventoryStatus


RangesStatusMarket Seg-

ment

RegionProduct GroupYear



TurnoverInventoryStatus


Customer

Type

OfficeProduct ClassQuarter

Customer

Group

Ware- houseSales OutletProduct NameMonth

SubgroupProduct No.

Sales, Order Qty, Material Cost per Unit, Discount, Percent Gross Margin,

Carrying Cost per Unit, Percent Material Cost per Sale, Percent Carrying Cost

per Sale, Sales per Customer, Percent Profit per Segment, Inventory Turnover.

Sample Meas-

ures

If you want to monitor customer service, you can expand your model to include indicators of service

quality. The new dimensions and categories might be encoded Reasons for Dissatisfaction or Causes

of Poor Quality Service.

Ensure that your source data provides the required measures, such as the number of complaints,

returns, and claims, or the dollar value of returns and claims.

Complete your model by incorporating response times, labor costs, time lost to service claims,

rework hours, scrap costs, or any other factor that affects service quality.

Diagnose Design ProblemsUse one or more of the following tools and techniques, to diagnose and resolve problems in your

model design:

❑ Show Scope

To see how your measures and levels are associated to their corresponding data sources by

allocation, direct, or indirect association, use the Show Scope command (Edit menu). If a level

has categories that are not unique, this will also be indicated onscreen. By default, levels with

uniqueness problems appear dark red.

❑ Show Count

To verify that you have maintained a 1:10 ratio or less between the categories in each level,

use the Show Counts command (Edit menu) . Lower ratios allow for efficient partitioning

and faster cube creation times in Transformer, and easier data exploration in PowerPlay.

❑ Show References

To confirm the origin of every data source column associated with your dimension map and

see how each is used, use the Show References command (Tools menu).

❑ Check Model

To ensure that the new data relates correctly to the dimension map as you add data sources,

and before you create categories, use the Check Model command (Tools menu) .



❑ Generate Selected Categories

To confirm how the categories in that specific data source relate to your model, use the Generate

Categories command (Run menu) on the selected data source . To prevent generation of

categories in specific levels or entire dimensions, select the Prohibit Automatic Generation of

Categories box on the Level or Dimension property sheets (General tab).

❑ Create Selected Cubes

During the prototyping stage, you may want to create only certain cubes. You can enable or

disable cube creation in one of three ways:

● change the Cube Creation cube option on the Processing tab of the PowerCube or Cube

Group property sheet

● use the Create Selected PowerCube command (Run menu)

● use the Model Definition Language (MDL) option CreateFromCubes

❑ Check Cube Status

You can check the status of a cube at any time without opening it by using the PowerCube

Status command (Tools menu). You can check the status for all the cubes that are defined for

a model at the same time. If you have more than one cube, you can apply a filter to monitor

the status of specific cubes as they are being created.

❑ Consult the Troubleshooting Help

In addition to the Troubleshooting chapter of this document, online help is available from

several Transformer error message windows, to assist you in overcoming production problems.

Review the Resulting Reports with UsersAsk for feedback from representative users, directing their attention by asking open-ended questions.

Review their answers with a business analysis, and keep a list of follow-up questions:

● Does each dimension level generate valid data, with measures that are properly associated or

coordinated, regardless of the data source? (Try to spot measures that don't roll up as expected,

or that aren't additive in every dimension.)

● Are ranges or qualitative values coded realistically? Are the values for key indicators consistent,

or is the integrity of the underlying data suspect? (In some cases, you may need to add other

measures that substitute average figures, or industry standards, for non-available or non-con-

tinuous values.)

● Is the data at some of the lower drill-down levels too sparse to be useful? Should the model be

redesigned, or should drill-down targets be added? (Consider expressing some values as ranges

rather than absolutes, to create useful groupings such as responsiveness or rates of return, for

example, or to hide sensitive details, such as salaries.)

● Could there be better coordination of the data flowing from different databases, models, and

reports, perhaps by using normalized measures, to ensure that computer resources aren't

overburdened?



● Has anyone developed a calculated column or exception dimension that could be added to the

standard reports, for the benefit of all?

Maintain regular contact with your report users, and incorporate their feedback into your system

enhancement plans. As you enhance or change your model, remember to use the label and

description fields for each dimension, level, and measure, so that cubes created from your model

are clear and intuitive.

Recommendation - Manually Partition Cubes that do not Sup-port Auto-partitioning

For PowerCubes containing more than 250,000 records, we recommend that you use partitioning

to pre-summarize the most frequently queried data into smaller subordinate groupings, to achieve

satisfactory runtime performance. Although there is an auto-partitioning capability, partitioning

will be done manually in the following circumstances:

● Any cube in the model uses externally rolled-up or before-rollup calculated measures.

● Any cube in the model has its Consolidation property set to No.

● Any cube in the model has a custom view with cloaking enabled for a primary drill category.

● Any cube in the model uses optimization methods other than Auto-partition or Default.

● Any cube in the model is set to use incremental update.

Process

❑ Begin by letting Transformer auto-partition your cube, adjusting the following settings:

● Estimated Number of Consolidated Records. Consolidation combines records that contain

identical non-measure values, reducing cube size and improving runtime performance. Start

with the default, then check the log file (Step 2) for the real number.

● Faster Cube Creation/Faster Cube Access. Use the slider bar to center the setting.

● Desired Partition Size. Use this setting, which is based on the consolidated record count,

to select the number of categories in (size of) your partition. Default: 500,000.

● Maximum Number of Passes. Transformer uses one pass for each partition level. If you

decrease the desired partition size, the number of passes (and partition levels) increases,

lengthening cube creation time. We recommended a 5-pass maximum.

❑ To find out how many consolidated records are in your PowerCube, build it once with the

default settings, and consult the Transformer log file. Look at the difference between the row

and category counts in the Start and End Count and Consolidation lines. In this example, no

consolidation was done on Pass 0; it was done on passes 1 through 4:

--- Performing Pass 0 with 22770681 rows and 8708 categoriesremaining.

Selected Dimension 3 for next pass of partitioning.



Sorting the work file.

Counting category hits.

End sorting 22770681 records.

Start Count and Consolidation with 22770681 rows and8708 categories remaining.

End Count and Consolidation with 22770681 rows and 8708categories remaining.

Start Write leaving 8708 categories remaining.

Updating the PowerCube data.

...

Performing DataBase Commit at record number 22770682.

End Write leaving 8708 categories remaining.








...



---Performing Pass 2 with 15522151 rows and 8708 categories remaining.








...













...











...






...



❑ Observe that the auto-partitioning algorithm partitioned dimensions 3, 11, 0, and 1, making

2 passes on time dimension 0 to get more consolidation on the second pass. The next dimension

(first non-time dimension) corresponds to number 1 in the log file, and so on. To assess the

effectiveness, compare Pass 0 to Pass 5 (or the last Pass listed in your log file):

--- PerformingPass 0 with 22770681 rows and 8708 categories remaining.


Auto-partitioning consolidated the original row count (22,770,681) to 7,312,012 and did some

hierarchical consolidation of the original category count (8,708) to 8,542.

❑ Note any dimensions that must be excluded from partitioning, and then select that option on

the General tab of the property sheet for the dimension. Typical reasons are as follows:

● a dimension is large and flat

● dimension categories change frequently (such as the time dimension)

● a dimension is frequently reported on at the lowest (leaf) level



● a dimension contains alternate drill-down paths

If you exclude your largest dimensions, you can set a smaller partition size, achieving faster

cube builds and summary querying. Also, by not partitioning on the detail dimensions, you

avoid the performance costs of forcing your low-level queries to span multiple partitions.

❑ Analyze the log file to see how the number of records and categories decreases from pass to

pass. If there is no decrease in the number of records on the last pass, reduce the Maximum

Number of Passes by one, or increase the desired partition size.

❑ Fine-tune the auto-partitioning settings as required, to achieve optimal results. Try increasing

the partition size from the default to 1,000,000 or 1,500,000. The partitioning passes will cease

when the number of records in the summary partition drops below this partition size. Test and

either increase the number again, or decrease it until performance is acceptable. If both the

summary and Level 1 partitions have the same number of records, indicating that the summary

partition was not consolidated, try increasing the Maximum Number of Passes.

❑ Check the partitioning information by right-clicking the cube in the PowerCubes list and

selecting PowerCube Partition Status. The window shows each applied partition level (beginning

with 0), its dimension name, the category code, and its record count. If you still need to improve

performance, perhaps for specific reports, continue partitioning manually.

Example

You decide to manually partition a flat-dimensioned model consisting of Product IDs, Customer

Numbers, and Countries. The ideal parent:child ratio of 1:10 is exceeded: there is only one parent

for each of the large child category levels (Product Brand, Customer Type, and Region).

You know your customers need summary reports on categories from the Product and Customer

dimensions, so you partition at a higher level in those dimensions. You specify which levels and

categories are included in the partitioning and set the Maximum Number of Passes on the Auto-

partition tab to that number of partition levels plus one. You test, and then try using various parti-

tioning strategies for the Products dimension.

At first, you assign a single partition to the entire Product Type level, with this result:

Level 0, All dimensions, <Summary> category - recordcount of 237

Level 1, Line dimension, <Type 1> category - recordcount of 31

Level 1, Line dimension, < Type 2> category - recordcount of 181

Level 1, Line dimension, < Type 3> category - recordcount of 25

However, drill-down performance is only acceptable when querying Types 1 and 2. It is unacceptably

slow for Type 3, because there are just too many records in this partition.

To more evenly distribute the total number of categories, you navigate to the General tab of each

Product Type category property sheet and assign the same partition number (1) to all Type 3 child

categories, and to all Type 1 and 2 categories at the parent level.

You rebuild the cube, check the partition status, and confirm that the Product Type category counts

are better balanced, with no partitions containing more than 87 records. However, you note that



some Region categories contain a large number of child records. You decide to add individual

Country categories to the same partition as that assigned to the parent, to further optimize parti-

tioning. You then recheck query performance, but conclude it is unsatisfactory.

You contact Cognos Software Services, first gathering up the following information, to facilitate

timely investigation of your problem:

● The Transformer log file from the PowerCube build, with information about which dimensions

were involved in the partitioning, and partition consolidation details.

● The populated Transformer model (.mdl or .py?) and PowerCube (.mdc), so Cognos Support

can confirm the status and record distribution of all partitioned categories.

● Information about which dimensions are most queried at runtime, and how deep those queries

tend to be. That is, do your users mostly query the parent or child levels?

● Detailed information about cases when query performance was below expectations, such as

typical reports and OLAP/explorer activity, especially regarding nesting or alternate drills.

Recommendation - Test Alternatives for Improving Build Timesfor Large PowerCubes

We recommend that for large PowerCubes (greater than 500,000 categories, with parent:child ratios

that deviate significantly from the ideal ratio of 1:10), you experiment with alternatives for

improving cube build times. This section discusses various hardware upgrades, memory adjustments,

model optimization techniques, and processing options, with case study examples of the efficiency

gains that you can expect with each.

Example - Explore the Test ModelA test model is used to troubleshoot problems and arrive at recommendations for optimizing build

times in a typical production environment.

Test Model Description

Total number of categories: 492,152

Number of non-measure dimensions: 5

Number of measures: 5 (2 calculated; 2 after-rollup calculated)

Source data format: ASCII ('~' delimited)

Number of source files: 9 (6 structural, 3 transactional); multiprocessing enabled for 4 largest

Number of transaction input records: 50 million

Size of all source files: 2.28 GB

Options: crosstab-caching and auto-partitioning enabled; 5 partition passes; 500,000 size limit

The following grid shows the model details, as laid out on the Transformer (Windows) interface.



Country of export(286) (2)

Country of origin(286) (2)

HR code(491,376) (2)

Date (197) (2)Transportation (7)(2)

Region 10 categoriesRegion 10 categoriesHR Code 1 1471

categories

Year 15 categoriesTransportation 5

categories

Country 274 categor-

ies

Country 274 categor-

ies

HR Code 2 19036

categories

Month 180 categor-

ies

HR Code 3 75916

categories

HR Code 4 123856

categories

HR Code 5 271095

categories

Gains from Multiprocessing on a Faster Transformer ServerBy upgrading to a multi-CPU Transformer server and running multiple instances in parallel, you

can significantly decrease cube build time. We recommend that you try the following strategies:

● Give each Transformer process its own CPU; that is, enable multiprocessing and assign each

Transformer instance two dedicated CPUs. Because each instance uses system resources

independently, ensure that each one has sufficient memory, I/O bandwidth, and disk space.

● Provide each Transformer instance with its own configuration files.

● Do not share the locations of the DataWorkDirectory and ModelWorkDirectory among multiple

instances.

● Add an ampersand (&) to the end of the UNIX command line to begin your first process in

background mode: rsserver -mmodel.mdl & When control returns to the command prompt,

initiate a second rsserver command in parallel.

● To continue Transformer on UNIX processing after session log off, type nohup rsserver

-mmodel.mdl

Typical Results

Typical optimization efficiencies were obtained with models similar to the test one.

● Adding a second CPU to the Windows server reduced data read times for the ASCII version of

the model by 25%. For the RDBMS version, data read times were reduced by 50%.

● Upgrading to a 2.5-times faster dual CPU on Windows improved the build time by 30%. A

50% improvement resulted on upgrading to a faster quad CPU on UNIX.



● Upgrading from dual 300 MHz UNIX processors with 1 GB of RAM to 8-750 MHz processors

with 16 GB of RAM significantly shortened every phase in the build process. Total cube build

time dropped from approximately 5.5 hours to less than 1.5 hours.

● A follow-up test on a model with 250 million transaction records showed even more dramatic

gains. Build time dropped from more than 35 hours to less than 6.5 hours, after upgrading

from a dual to a quad CPU, running 3-times faster, with 4-times the RAM.

At some point, hardware upgrades alone may not solve your slow build problems. A comparison

test was run on two models using the fastest, most powerful CPUs available. The first had 32 source

files with 500 million transaction records. The second had 68 source files with a billion transaction

records. The 2.2 GB cube took more than 12 hours to build; the 4.6 GB cube, more than 26 hours,

an unacceptably long time in a production environment. If you are experiencing similar problems,

we recommend that you consider other optimization techniques.

Gains from Adjusting the Memory AllocationWe recommend that you have a server dedicated to building cubes. The server should have sufficient

memory to handle all running application requests, and an operating system disk cache that can

grow as large as required during the cube build. Excessive paging occurs if there is not enough

physical memory available, and this can severely impact overall cube build times.

After you optimize the memory settings for your server to build cubes faster, we recommend that

you track virtual and working memory use, to find the bottlenecks as Transformer processes the

categories in your largest, slowest cubes.

Typically, total addressable (virtual) memory usage climbs rapidly during category generation, and

remains relatively constant during all cube build phases (read data, update metadata and update

cube). Working memory, or that portion of the physical memory that is used by Transformer, also

rises quickly and remains high until the cube update phase, when it is freed for use by the operating

system disk cache.

In the first phase, the more categories that there are, the more memory that is required. Although

models differ, working memory usage is typically 500-1,500 bytes per category. Paging (swap file

use) is needed if the space allocated to working memory is insufficient and cannot grow to the

amount required to process the model.

In a comparison test, the system was first tuned to allow Transformer to use all available system

memory (512 MB). This was compared to allowing Transformer only 128 MB of working memory,

well below the 200 MB requirement established in the previous test. The build process took nearly

50% longer, slowing down during both the data read and metadata update phases.

When page file activity was checked in the 512 MB system, working memory usage dropped at that

point where page file usage spiked. In the 128 MB test, working memory usage was only 25-50%,

forcing page file usage to remain near capacity through all phases of the build. Excessive paging

severely impacted the time taken to build the 128 MB test cube.



Gains from Reconfiguring the Hard DriveYou can achieve additional gains by reconfiguring the hard drive of your build server to optimize

I/O processing. We recommend that you allocate at least three physical disk drives to your

Transformer system, subdividing the processing as follows:

● Drive 1: operating system and Transformer executable

● Drive 2: Transformer DataWorkDirectory

● Drive 3: Transformer ModelWorkDirectory and CubeSaveDirectory

The log file for this configuration reads as follows, where 1, 2, and 3 represent drives c, d, and e:

LogFileDirectory=c:\transformer\logs

ModelSaveDirectory=c:\transformer\models\

DataSourceDirectory=c:\transformer\data\

CubeSaveDirectory=e:\transformer\cubes\

DataWorkDirectory=d:\temp\

ModelWorkDirectory=e:\temp\

You may also want to configure the build server to use striping (RAID level 0) rather than mirroring,

assuming that you have a backup for use if the system crashes during a production build. The build

time is reduced because data is distributed among nonredundant disk drives.

Gains from Allocating Sufficient Space for the Temporary FilesDuring the three cube build phases, you need varying amounts of space for the temporary files.

Data Read Phase

Transformer reads the source data into a temporary work file based on the model structure. Insuf-

ficient disk space and database connectivity can cause problems.

Metadata Update Phase

Transformer compares the contents of the temporary work file to the categories in the model, to

see which ones go into the cube, and creates a copy of the file. The original work file is only deleted

after processing is complete and eligible categories are inserted into the cube. Insufficient disk space

and lack of memory can cause problems.

Data Update Phase

Before the data points in the temporary work file are inserted into a partitioned cube, the data must

be sorted, requiring several passes through the temporary file. The most common issue during this

phase is low system memory.

You can estimate the amount of space to allocate for the temporary files. If {([(x2*4)+(x3*4)+

(x4*9)+8]*x5)/1024/1024}*2 is greater than 1907, the space calculation is that number plus x6.

If the calculated result is less than or equal to 1907, the required space is that number (x6 not

added).

In our test model, with the input values shown, the calculated space requirement was 7047 MB:

● x2 = 5; the total number of dimensions



● x3 = 0; the number of dimension views associated with the PowerCube

● x4 = 3: the number of regular measures (calculated measures are not counted)

● x5 = 49000000; the number of input records for all transactional data sources

● x6 = 1907; the WorkFileMaxSize trnsfrmr.ini setting divided by 1024 * 1024 (MB)

To this number, you must add sufficient space for the PowerCube and the model checkpoint files:

7 GB + 20% for the former, and a number equal to the model working space for the latter.

We recommend that you point the Sort directory to the same spot as the Data Temporary file, with

all other Transformer directory locations pointing to another drive. You should then use a system

performance monitor during the cube build, to check the amount of available disk space.

Gains from Optimizing the Windows EnvironmentWe recommend that you adjust these settings to shorten build times on Windows systems.

WriteCacheSize

Where possible, allow the memory used by the disk cache to expand up to the final size of the

PowerCube. You can change this setting using Series 7 Configuration Manager. Under Services -

PowerPlay Data Services - Cache, incrementally raise the default value (8192 or 8 MB) in increments

of 1024 (1 MB), up to 32768 (32 MB) or 65536 (64 MB). Do not exceed 100 MB because this can

degrade performance.

Note: The WriteCacheSize is set for the Transformer application as a whole. In Series 7, you can

use the PowerPlay Enterprise Server Administration tool to set the read cache on a cube-by-cube

basis, impacting all Web queries that target that specific cube.

SortMemory

You can raise the amount of physical memory available for sorting data during the consolidation

and auto-partitioning process, in 2 KB blocks. The default is 512, which provides 512 x 2 KB or

1 MB of memory. We recommend that you change the setting in Series 7 Configuration Manager,

under Services - UDA - General, to 5120 (or 10 MB).

TEMPFILEDIRS

You can change the location of the temporary files that are created whenever sorting is done. We

recommend that you specify multiple directories, separating each with semicolons. You change this

setting in Series 7 Configuration Manager, under Services - UDA - General.

MaxTransactionNum

You can change the setting that limits the number of records being processed in temporary files

before a checkpoint is inserted and records are committed to the PowerCube. Change the default

setting (MaxTransactionNum=500000) on the General tab of the Series 7 Transformer Preferencesdialog box. For example, if you see the TR0112 error during a cube build, try lowering the Max-

TransactionNum so that it commits more frequently. This will free up space. To reduce build times,

try increasing the value, perhaps to 800000.



Gains from Optimizing the UNIX EnvironmentConsider changing the UNIX equivalents to the Windows settings, to shorten build times on UNIX.

PPDS_WRITE_MEMORY

This is similar to WriteCacheSize on Windows, except that the default value for this environment

variable is 32768 (32 MB). Doubling the values to 65536 (64 MB) or tripling them to 98304 (96

MB) is recommended to optimize larger systems.

SORTMEMORY

This environment variable is equivalent to SortMemory on Windows.

TEMPFILEDIRS

This variable is equivalent to the Windows setting of the same name.

MaxTransactionNum

This variable is similar to the Windows setting of the same name except that the default can be

changed in a trnsfrmr.rc file or set in a Preferences file.

Ulimit

We recommend that you change this environment variable to unlimited. Typically, UNIX systems

specify a particular setting, such as 67 MB on a 2 GB server, so that system resources are effectively

shared among competing processes. However, Transformer requires as much physical memory as

possible for efficient cube builds.

You can determine the ulimit assigned to rsserver (the Transformer on UNIX executable) by typing

on the command line: ulimit -a The following information appears:

time(seconds): unlimited

file(blocks): unlimited

data(kbytes): 65536

stack(kbytes): 8192

memory(kbytes): unlimited

coredump(blocks): 4194303

nofiles(descriptors): 1024

Note: For best results on HP-UX, set the memory (kbytes) option to unlimited. For information

about other UNIX platforms, contact the system administrator or operating system documentation

to determine how to tune your kernel settings to allocate sufficient memory for Transformer.

Gains from Redistributing FilesThe following settings can be globally specified by means of preference files:

● ModelWorkDirectory=<path>

Specifies where Transformer creates temporary files while you work on your model. The tem-

porary file can be used to recover a suspended model at strategic checkpoints should a severe



error occur during cube creation. This file has the extension .qy? and the default path is the

value of the ModelSaveDirectory setting.

● DataWorkDirectory=<path1;path2;...>

Specifies where Transformer creates temporary work files while building cubes. By using multiple

drives, you can work around size limitations imposed by your operating system. As Transformer

builds cubes, it writes temporary files to the specified drives or directories. The files are then

concatenated into one logical file, regardless of which drive they are in. The location of these

files is determined by the list of paths that you specify. The default path is the value of the

CubeSaveDirectory setting.

● DataSourceDirectory=<path>

For data source files other than IQD files, this setting specifies where Transformer searches for

files. The default path is the current working directory.

● CubeSaveDirectory=<path>

Specifies where Transformer saves cubes. The default path is ModelSaveDirectory.

● ModelSaveDirectory=<path>

Specifies where Transformer saves models. The default path is the current working directory.

You can also specify preference settings on the command line. These settings override or take pre-

cedence over all other settings, including environment settings defined in the rsserver.sh file, or the

environment variables TMPDIR, TEMP, and TMP as defined by the operating system. If multiple

variables are defined, Transformer uses the first one in the list.

The Windows command line syntax for specifying global preferences is

trnsfrmr -n -fc:\preferences.prf model.mdl

The equivalent command line syntax for UNIX is

rsserver -F preferences.rc -mmodel.mdl

Gains from Optimizing Gateway SettingsTo further shorten the data read phase, you can change the database-specific settings found in the

gateway INI files included in the cern Series 7 Transformer installation directory. The file names

are cogdm*.ini, where the asterisk represents a specific database version. For example, Oracle uses

cogdmor. This file contains the following settings:

Fetch Number of Rows

Increasing the number of rows to fetch in each fetch operation can improve performance on some

systems. Although the current limit for this number is 32767, numbers larger than the default (100)

may degrade performance on some systems.

Fetch Buffer Size

Increasing the size of buffer used during fetch operations from the default (2048 bytes) can improve

performance on some systems. Note that where both entries have been reset, the row setting takes

precedence over the buffer size setting.



Analyze Processing Bottlenecks Using the Transformer Log FileWe recommend that you examine the Transformer log file if processing bottlenecks occur during

cube builds.

Steps

1. Import these delimited-field data files into a spreadsheet, such as Microsoft Excel:

LogFileDirectory=f: \logfiles

ModelSaveDirectory= \models\

DataSourceDirectory= \data\

CubeSaveDirectory= \cubes\

DataWorkDirectory=F:\datawork\

ModelWorkDirectory= \modelwork\

2. In the resulting log file import, select the header of the E column and, if you are using Microsoft

Excel, from the Data menu, click Filter and select the AutoFilter option.

3. In the list, select either (NonBlanks) or one of the three phases in the process. The spreadsheet

now shows only the timing information for your selection.

4. Select several cells in the F column and select the Sum command, to add the timing values for

your selected cells, as displayed in the lower toolbar.

5. Repeat, to examine each subphase. These appear in distinct blocks and each completes before

the next subphase begins.

Data Read

INITIALIZING CATEGORIES

OPEN DATA SOURCE

READ DATA SOURCE

MARKING CATEGORIES USED

Metadata Update

SORTING

UPDATE CATEGORY AND PROCESS WORK FILE

METADATA

Data Update

CUBE UPDATE

CUBE COMMIT

Note: If the timing shown for TOTAL TIME (CREATE CUBE) is different from the timing for

the separate phases, you can adjust for the time difference in the Cube Update phase.



6. Assess how much time is spent on each phase. If you notice that, over time, cube builds are

taking longer to complete even though the data set is relatively constant, compare each phase

in the log files, to see where increases in build times are occurring.

Keeping Model and Cube Sizes Within Practical LimitsWe recommend that you limit the size of your Series 7 models to 2 million categories each. Pay

particular attention to bloating due to excessive use of labels, descriptions, short names, category

codes, and custom views. Metadata associated with your structural data source dimensions, levels,

and categories can contribute significantly to overall storage requirements.

File size

The old 2 GB limit does not apply to Series 7 Version 4 and subsequent releases.

Virtual memory

Models are held in allocated memory limited by available address space.

Physical memory

Performance is severely impacted if a model is larger than the available physical memory.

Performance Improvement Case StudiesThree case studies illustrate how build times can be shortened by using one or more of the available

optimization techniques. All timing results were generated using actual production data, in an

isolated laboratory, without external influences. No other applications were active.

Recommendation - Uncloak the Primary Drill CategoryIn this example, the production cube was taking too long to build, primarily due to the large number

of transactional input records. When Check Model was run, error TR2757 indicated that one or

more cubes used a dimension view with a cloaked primary drill-down path. This meant that auto-

partitioning could not be enabled. The fix involved uncloaking the drill category, which enabled

auto-partitioning.

Model Design

Number of categories: 546,391

Number of dimension:9 (measures not counted as a dimension)

Number of measures: 10; four calculated:

"Category A" * 100 / "Category B"

"Category C" * 100 / "Category B"

"Category D" * 100 / "Category B"

"Category E" * 100 / "Category B"

Source data format: ASCII (',' delimited)



Number of source files: 23 (13 structural, 10 transactional); multiprocessing enabled

Approximate number of input records: 3 million

Size of all source files: 1.28GB

Cube Build Times

Build time after the fixBuild time before the fixLogged build phase

28 minutes28 minutesREAD DATA SOURCE

4 minutes2 hours, 10 minutesMETADATA

4 minutes4 hours, 49 minutesCUBE COMMIT

41 minutes7 hours, 41 minutesTOTAL TIME (CREATE CUBE)

Recommendation - Enable MultiprocessingIn this example, many cubes had to be built within a specified time frame. The most feasible solution

was to enable multiple instances of Transformer to run in parallel. The system was redesigned to

have an 8-CPU server build all the cubes at once by enabling multiprocessing.

Although individual cube build times proved marginally longer after the fix, the total create time

was reduced by 2 hours and 23 minutes, to 12 hours and 2 minutes (the time taken for the model

that was slowest to build, C).

Model A Design


Number of dimensions: 5 (measures not counted as a dimension)

Number of measures: 5; two calculated:

("Category A" - "Category B") / 100

"Category A" - "Category B"


Number of source files:6 (6 structural, 3 transactional); multiprocessing enabled for 4 largest


Size of all source files: 2.28GB

Model B Design



Number of measures: 15




Number of source files: 6 (5 structural, 1 transactional); multiprocessing enabled for 5 largest


Size of all source files: 224 MB

Model C Design



Number of measures: 10 (four calculated)


Number of source files: 23 (13 structural, 10 transactional); multiprocessing enabled for all


Size of all source files: 1.28 GB

Model D Design



Number of measures: 4 (one calculated)


Number of source files: 14 (11 structural, 4 transactional); multiprocessing enabled for 3


Size of all source files:569 MB

Cube Build Times

Build time after the fixBuild time before the fixLogged build phase

2 hours, 41 minutes2 hours, 24 minutesModel A

14 minutes12 minutesModel B

12 hours, 2 minutes11 hours, 31 minutesModel C

19 minutes16 minutesModel D

12 hours, 2 minutes14 hours, 25 minutesTOTAL TIME (CREATE CUBE)

Recommendation - Use Time-based PartitioningIn this example, the production team ran a test to compare the relative effectiveness of two optim-

ization strategies: time-based partitioning versus incremental cube updates. With the incremental

cube update technique, several data updates had to run because auto-partitioning was not performed.



The overall cube build time was very slow. With time-based partitioning, the cube built much faster

because the lengthy data update phase was no longer required.

Model Design



Number of measures: 28 (eight calculated)

Source data format: IQD files

Number of source files: 9; multiprocessing enabled for all

Approximate number of input records: 2 million for the incremental update test

Approximate number of input records: 1.1 million for the time-based partitioned cube test

Cube Build Times

Time-based partitioningIncremental updateLogged build Phase

7 minutes23 minutesREAD DATA SOURCE

22 seconds20 minutesMETADATA

5 seconds6 minutesCUBE COMMIT

N/A11 hoursDATA UPDATE

14 minutes12 hours, 5 minutesTOTAL TIME (CREATE CUBE)

Note: The number of data source records differed between the two tests. However, gains were

several orders of magnitude, proving that time-based partitioning is by far the better strategy.

Recommendation - Resolve Uniqueness Problems in the SourceBefore Moving PowerCubes into IBM Cognos 8

To avoid level uniqueness problems, it is your responsibility (as the Transformer modeler) to ensure

that no two categories in a level represent identically-named distinct categories, such as cities with

the same name in two or more regions.

Flagging categories as unique by selecting the appropriate Series 7 Transformer check box does not

automatically make them so. You are merely specifying that this property is true for the source data

used in your model.

If you indicate that the categories in a level are unique when they are not, invalid results will occur,

because Transformer cannot reliably associate the non-unique values to the appropriate categories.

When the source value is read, Transformer places the value in the first category identified as being

associated with that value. Duplicate categories in the level are renamed using the tilde character

(~), making them non-numeric.



Problems with uniqueness will present additional problems in IBM Cognos 8, especially with drill-

through operations, because Member Unique Names (MUNs) must be mappable to specific categor-

ies, to avoid unexpected results.

It is possible that your source data might contain columns that populate levels you thought were

unique that are in fact not unique. In such cases, a message informs you of the potential problem

when you attempt to generate categories, but only if the data source contains all of the columns

required for the levels in question. If categories for some levels have values derived from other

transactional data sources, uniqueness conflicts may arise but be undetected. Again, if you selected

the Series 7 Transformer optimization setting to maximize query speed, the model is not checked

for uniqueness conflicts.

Removing Missing Values that Cause Uniqueness ProblemsIt is possible to declare that a level contains unique source values when it truly does, and still

encounter uniqueness problems. This arises when you create an "unbalanced" dimension in which

more than one category value in a level is missing.

If the source data contains missing values for a measure, Series 7 Transformer writes the value

"na"or 0 to the PowerCube. To avoid these values being retained and causing uniqueness problems,

we recommend you do one of the following:

● Clean your source data so records with missing values are excluded, or so missing values are

eliminated at the source.

● Omit the measure from the PowerCube if the measure has missing values that are not required.

● Allocate the measure values to lower-level categories in the dimension.

For example, suppose you create a Regions dimension that contains the levels Country, State, and

City. It is reasonable to assume that the State level is unique. However, if some countries do not

have states, the State column for those countries will be blank and non-unique values will be

introduced (in the form of missing values). To avoid this kind of uniqueness violation, we recommend

that you create a subdimension for that portion of the dimension that does contain State values,

and designate that State level as unique. You can then omit the portion of the dimension with the

missing States.

Resolving Uniqueness Problems in the SourceIf you are using Impromptu IQD files as your data source, you can use Impromptu to make the

categories unique. To do so, use Impromptu to create calculated columns that combine the original

non-unique source values with a qualifying code or identifier. You can use the resulting unique

column to provide unique category values to Transformer.

If you are using a data source other than IQD, you must use your IBM Cognos or third-party query

tool to render the level values unique, or you must refine the source data so that it does provide

unique values. Then make sure that the appropriate settings are specified in the appropriate property

sheets of your Series 7 Transformer model, so data is correctly mapped into IBM Cognos 8.



Retaining the Uniqueness Designation when Categories are MovedYou can prevent model errors due to changes in your data source by designating a unique level.

This tells Transformer that categories in that level are identified by their source values alone, without

reference to their ancestors. However, when you move a category from a unique level, a uniqueness

violation is reported during category generation because the moved category now appears in a dif-

ferent context.

To avoid having to manually restructure the categories in a unique level to conform to the changed

ancestor data, you can specify that such changes be treated as unique moves. Measure values, even

those accumulated under the old structure, are thereafter rolled up the new path to the moved cat-

egories. Unless you make special adjustments in your model (such as implementing calculated

columns to demarcate the move date), historical values move to the new path, and are no longer

associated with the old one.

Note: In certain circumstances, Unique Move is not supported in time-based partitioned cubes.

Preventing New (Non-unique) Categories from Being AddedWhenever you generate categories for a model, Transformer detects new categories as it reads data

values from the sources in the Data Sources list. When this happens, new categories are typically

added to the model.

If this is undesirable in your current situation, you can prevent Series 7 Transformer (Windows)

from adding new categories to a dimension by selecting the Prohibit Automatic Creation of New

Categories check box on the property sheet for the applicable dimension. From the command line,

you can use the -o option.



Appendix E: General Troubleshooting

This section describes how to work around various problems you may encounter when working

with products in the IBM Cognos Business Intelligence group:

● "With the IBM Cognos 8 Web Studios" (p. 283)

● "With Transformer on Windows" (p. 285)

● "With Transformer on UNIX" (p. 299)

● "With PowerPlay Client Applications" (p. 300)

With the IBM Cognos 8 Web Studios

Choosing Only One Language When Installing IBM Cognos 8 OLAP Modelingand Mobile Analysis

The IBM Cognos 8 BI OLAP Modeling and Mobile Analysis packages are delivered on multiple

CDs. Each CD is organized into two or more top-level language folders such as English, French,

and Japanese. When installing these products, you must install them from one of the available lan-

guage folders.

The language that you choose becomes the language of the user interface of the product. However,

the product can process data in any of the supported languages.

Providing Stricter Access Controls for PowerCubes Used in IBM Cognos 8When a Series 7 namespace is used to secure PowerCube content, access to information is controlled

by means of user class views embedded in the PowerCube model. After deployment to IBM Cognos 8,

these PowerCube files reside in the target directory (deployed environment), with no additional

controls on user access.

To provide enhanced data security, we recommend that file system permissions be applied to the

directory containing the deployed PowerCubes. For example, grant Read access for the directory

to the domain account used to start the IBM Cognos 8 service. Grant Read/Write access for the

directory to any user accounts that are permitted to add or remove cubes found in the directory.

Not Yet Optimized IBM Cognos PowerCubes May Open Slowly in IBM Cognos 8If your PowerCubes take too long to open in the IBM Cognos 8 Web studios, we recommend that

you run a command line utility named pcoptimizer, supplied with IBM Cognos 8, to improve

runtime performance. This optimization utility is suitable for older PowerCubes when the model

no longer exists or the data used to build the PowerCube is no longer available.

In Transformer Version 7.3 Maintenance Release 2 and subsequent releases, you have the option

of making this optimization an automatic part of the build process, by adding an entry to your


trnsfrmr.ini file or by creating the required environment variable on UNIX. If you use this method,

the optimization process also enables drill-through access on all PowerCube measures by default.

There is an expected, but slight, increase in PowerCube size and build times. However, in most

production environments, runtime PowerCube performance improves significantly.

Steps to Use the Command Line Utility (IBM Cognos 8)

1. Back up your target PowerCube as a precaution, and then navigate to the IBM_Cognos_8_

installation_location/bin directory.

2. On Windows, open a command line window and run PCOptimizer.exe.

3. On UNIX, enter the following line to run the optimization command line utility:

pcoptimizer [-t] [-v] [-h] <cubename>

where <cubename> is the fully qualified PowerCube or time-based partitioned control cube

name with the .mdc extension, if the PowerCube resides in the same location as pcoptimizer.

Otherwise, <cubename> is the full path with the .mdc extension.

Note: This method only supports metadata extraction. To set up user-configurable drill-through,

you must use Transformer and specify automatic optimization. Wildcard character support is not

currently available. You must therefore invoke the utility once per PowerCube. If <cubename> is

not provided, the program enters an interactive mode, prompting you for a PowerCube name and

accepting keyboard input. The optional parameters are as follows:

● -t or test mode; it tests whether the extra metadata was extracted and loaded into the Power-

Cube. The return code indicates the status.

0 if the metadata was extracted and loaded

10 if the metadata was not loaded

20 if an error occurred while accessing the PowerCube

● -v or verbose mode; readable text is output to standard output (stdout), indicating what was

done, including any error messages. If running in interactive mode, -v is assumed. All text is

output in English only.

● -h for command-line help; if <cubename> is not provided, it prints the usage and options to

the screen.

Steps to Configure Automatic Optimization (Transformer Version 7.3)

1. On Windows, to enable automatic optimization for all PowerCube builds, locate the trnsfrmr.ini

file in the installation_location\cern\bin directory, open it in any text editor, and add the

following entry:

EnablePCOptimizer=1

Optimization is now turned ON for all PowerCube builds.

2. On the Windows command line, to optimize a specific PowerCube build, type the following:

trnsfrmr -DEnablePCOptimizer=1 -cm <model_path_and_filename>



3. On UNIX, to enable optimization for all PowerCube builds during the current session, use the

-D flag or define the environment variable EnablePCOptimizer in the trnsfrmr.rc file(s).

Tip: To disable the feature, change the value of your optimization setting to zero or remove

the definition.

4. After your PowerCubes build, you can open the relevant log files and confirm that the following

entries appear:

Start updating of 'cube_name' with optimized metadata.

. . .

End updating of 'cube_name' with optimized metadata.

This indicates successful extraction of the metadata needed to optimize the opening of each

PowerCube in IBM Cognos 8.

Note: There is no harm in optimizing a PowerCube that has already been optimized.

With Transformer on WindowsPersonal Transformer and the Mobile Analysis package of IBM Cognos 8 do not support the full

range of features carried by Series 7 Transformer. For more information, see the "Features and

Editions" topic in the Welcome chapter.

Preventing Errors When Model Calculations Use Double Quotation MarksIf you try to open an .mdl-format model containing calculations that include double quotation

marks, as might be used to create concatenated categories, you may get an error, even if you followed

the recommended practice of wrapping these calculations in single quotation marks.

This is because .mdl-format models do not support the use of single and double quotation marks

together, if the ObjectIDOutput flag is set to True, which is the default model creation setting.

To avoid this problem, you have two choices:

● You can open the INI file for the Transformer executable in any text editor, search for the line

ObjectIDOutput=1 and change the setting to 0, and then resave your problem model.

● If you must output object IDs, you can use your RDBMS or a tool such as IBM Cognos

Impromptu to perform the required calculations, and then import the data into your model.

Whichever strategy you choose, you can then open the .mdl or py?-format model without error.

Empty Results When Querying a Microsoft Access DatabaseIf a Microsoft Access query returns an empty result set, check the syntax of your "Where" filters

and "Like" expressions. In Transformer Version 7.3 and subsequent releases, ActiveX Data Object

(ADO) replaced the Data Access Object (DAO) layer. The ADO SQL syntax requires the use of a

different wildcard character: the percent sign (%), rather than an asterisk (*).

We recommend that you correct the syntax before running queries in Transformer. If you use the

Microsoft Access syntax at design time, you can continue to preview the results inside your Access

database.



Access Time Increases with Incremental UpdatesWhen you incrementally update a partitioned model, any new categories that are added from data

increments automatically get added to the partitions for their partition parents. As a result, cube

access time from PowerPlay may degrade as the cube grows. In addition, the model file can become

fragmented.

To remedy this, you should periodically rebuild your incrementally updated cubes from scratch

(using all data from all increments), at which time you can repartition the model. This allows you

to integrate new categories into your partitioning scheme.

Add New Partition Levels to Existing PowerCubesIf you are not using the incremental update feature to add data to your PowerCubes in increments,

you can add new partition levels and add new categories to existing partition levels at any time.

You cannot add new partition levels to a model used to create an existing incremental update

PowerCube. To change partitioning for the model, you must repartition the model and rebuild your

PowerCube from scratch, using all the data from all increments.

Retain Sort Order in Time-Based Partitioned CubesTransformer has recently been enhanced to better handle the sorting of merged categories in time-

based partitioned cubes. It now supports minor changes to the categories in a cube, from one update

to the next, with or without a Unique Move. In most such cases, the sort order of the categories

appears as expected when the child member cubes are viewed.

During updates and merges of time-based partitioned cubes, a few assumptions and constraints

must be kept in mind:

● The changes to categories must be relatively slow and orderly.

● There must be no significant changes to the category metadata in the overall model.

● An understanding of parent-child and sibling relationships must be available so that Transformer

can use that to help build the member cubes.

● Under certain circumstances, a unique move may affect the sorting or ordering of categories,

in unexpected ways.

● For allocation to work properly, you must select the Always Include option for the affected

categories.

Allocations Aren't Working CorrectlyAllocation is useful for measures that are not recorded at all levels in a dimension. For levels below

which measure values exist, you can allocate values from ancestor levels where the measure was

recorded.

There are several possibilities for measure allocation:

1. The measure was not recorded at any level in the dimension. It was recorded at some level in

another dimension. In this case, Transformer assumes that the measure is independent of the



dimension, and allocates the values for that measure down the categories in the dimension as

a constant. In the PowerPlay clients, the values for the measure do not change from level to

level as the user drills up and down through the dimension hierarchy.

2. The measure was recorded at a summary level in a dimension (for example, budgets are made

often only at the Quarter level in a date dimension). In this case, Transformer does not allocate

the values down to lower levels from the level at which the measure is recorded. For categories

below the level at which the measure is recorded, only zeros are displayed in the PowerPlay

clients.

3. The measure was recorded at the lowest level in a dimension. In this case, the measure is not

allocated in the dimension. However, it may be allocated in other dimensions (see case 1, above).

In cases 1 and 2 above, you can change the way in which measure values are allocated. In particular,

if your source data contains measure columns in some way proportionally related to the measure,

you can allocate proportionally based on those other measures.

Things to check if your allocations are not working as expected:

● The column properties for date values associated with the measure. For example, if sales forecasts

are made on a quarterly basis, ensure that the Degree of Detail for the date column in the data

source containing the sales forecast measure is set to Quarter. This ensures the measures are

recorded at the Quarter level in the model. You can then use other measures (such as last year's

actual sales) to allocate the quarterly forecasts to the month level.

● Ensure that you are not trying to allocate a calculated measure. Transformer will not allocate

calculated measures. However, you can use a calculated measure to allocate other measures

proportionally.

● Check that the measure is not referencing itself in some way in the allocation scheme. You

cannot allocate a measure by itself, or by a calculated measure that references the measure you

are trying to allocate.

● In dimensions with alternate drill down paths, do not attempt to allocate in any but the primary

drill down path. Transformer does not support allocation in alternate drill-down paths.

Alternate Drill-downs and Level UniquenessWhen you create an alternate drill-down, it is essential that alternate drill-down paths converge at

a level containing unique categories.

The concept of unique categories in a level is not the same as a unique key in a relational database.

Alternate Drill-downs and PartitioningWhen you partition your model, avoid partitioning on dimensions that contain alternate drill-down

paths. When you specify partition level numbers in a dimension, Transformer creates partitions for

the descendant categories. If you assign partition level numbers to categories in the primary drill-

down path of a dimension, then drill-down performance on the categories in the alternate drill-

down path may be poor. This is because the categories required for a summary value in the alternate

drill-down path will have to be read from multiple partitions.



Can't Join the Source Data Files in My Data Sources ListTransformer's focus is on associating data sources and values in their columns with the levels and

measures you define in your multidimensional model. Transformer processes each data source

independently, defining categories and/or generating measure values for PowerCubes. For this

reason, Transformer does not perform database-style joins between columns in your source data

files.

If you require joins between database tables, use a query tool (such as Impromptu) to create the

joins prior to adding data sources to the model.

Category Generation Is Too SlowThe amount of time Transformer takes to generate categories depends on several factors:

● the location from which Transformer is reading source data records. If you are accessing data

on a LAN or from a relational database, ensure the network and/or database load is not high

when you are building large models.

● the number of categories in the model. For very large models with tens of thousands of categories,

ensure the PC or server on which you are generating the categories has sufficient memory. You

can also adjust the amount of memory available to Transformer when building PowerCubes.

Dimension Contains Orphans Of CategoriesIn this case, Transformer has encountered a condition in which a transactional data source contains

orphaned data records. For example, suppose you use a Customers data source to populate your

Customers dimension and an Orders data source to provide measure values for your sales transac-

tions. You might encounter the condition in which a sale exists (in the Orders data source) for a

customer for whom no data record exists in the Customers data source. Transformer will include

customer information from the Orders data source, and fill in any blank levels in the Customers

dimension with "Orphans Of" categories.

To solve this problem, ensure that all references from data sources are to categories that have been

defined in the model.

Dimension Contains Unbalanced Hierarchy of CategoriesSee Errors: "Message TR2704" (p. 352).

Incremental Update Failed for One or More PowerCubesWhen you add data to PowerCubes in increments, Transformer maintains internal synchronization

indicators between the model and the PowerCube. Transformer does this to prevent a PowerCube

from being inadvertently updated with incorrect data.

There are three possible scenarios when you incrementally update a PowerCube:

1. Transformer issues a warning that it detected an invalid ID stamp on the PowerCube, and you

cancel the update.



2. Transformer issues a warning that it detected an invalid ID stamp on the PowerCube, and you

do not cancel the update.

3. The incremental update completely fails for some reason. For example, the MDC file may have

been locked when the update was attempted.

What to do for Case 1

Transformer sets the PowerCube status to invalid because the cube does not match what the model

expects. In this case, you use the Adopt feature to associate the model with the current PowerCube

if you know it is the correct one for the model. Alternatively, you can replace the cube (.mdc) file

with the correct one, or with one you restore from backup, and then use the Adopt feature to

associate that cube with the model.


Check the PowerCube carefully for potential problems. Because the model detected a synchronization

issue, it is possible you have incrementally updated the wrong version of the cube. This can mean

that you have added the wrong data to the wrong cube, or that you have added data to a cube twice

or more. If you detect this case, restore both the model and the associated PowerCube from backup

and repeat the update.

Note: When you back up an incrementally updated PowerCube, ensure that you make a backup

copy of the model as it was when that increment was processed. Restoring an old PowerCube with

a model that has processed more recent increments will lead to further synchronization errors.


You must restore both the model and its associated PowerCube from backup and repeat the update.

Transformer will not allow you to update an incrementally updated PowerCube with a Failed status.

Invalid Dates Generated in the Time DimensionIn the column for the date dimension, Transformer has encountered data values that it does not

recognize as valid date values. Possible occasions when this can occur are:

● your source data contains invalid date values (like 19963528)

● you haven't identified the date format of the date column correctly. See "Define the Date Input

Format" (p. 76).

Level Uniqueness ProblemsYou have encountered a problem relating to level uniqueness. This can occur when:

● you disable some of Transformer's error checking options so that it no longer verifies unique

levels don't contain non-unique categories.

● you use multiple data sources in a model and you don't define levels associated with columns

from multiple data sources as unique.

● you create an alternate drill-down structure that requires a unique converging level, and

Transformer asks you to verify the uniqueness property before proceeding.



● you specify that a level is unique and Transformer detects that it is not.

When Transformer encounters one of these conditions, it issues one of the following messages:

● TR1503: Choosing 'Maximize Query Speed' option will increase processing speed significantly

in some cases but Transformer will no longer detect multiple instances of source values in levels

marked as Unique.

● TR2312: You made level <level_name> into a convergence level that connects two or more

alternate drill-down paths. Categories in convergence level must have unique source values.

Can there be two categories in this level with the same source value?

● TR2313: You are creating a level as a convergence level that connects two or more alternate

drill-down paths. Categories in convergence levels must have unique source values. Is this level

unique by this definition?

● TR2314: This is not an unique level.

● This text contains proprietary information which is protected by copyright. All rights are

reserved. No part of this document may be photocopied, reproduced, or translated into another

language without the prior written consent of IBM (formerly Cognos Incorporated).

● TR2317: The level <level_name> is designated as unique. Source value <value> was used in an

attempt to create a category in the path <path2>. <value> already exists in level <level_name>

in the path <path1>.

● TR2318: Transformer has detected <n> attempts to create a category in more than one path.

Refer to the online help for a detailed explanation of level uniqueness.

● TR2319: A convergence level in an alternate drill-down path must be unique.

What Uniqueness Means

The concept of uniqueness used with dimension levels in Transformer is not the same as the concept

of uniqueness used in relational databases. A column that is a unique key in a database table does

not necessarily provide unique category values for a level in a Transformer model.

Level uniqueness in Transformer means that the level contains no more than one instance of each

category source value. In practical terms, this mean no two categories in the level can represent

● identically-named distinct categories, such as cities with the same name in two or more regions.

For example, in a Regions dimension with levels Country, State, and City, there may be cities

named Burlington in both Massachusetts and Vermont. If so, the City level is not unique.

● a "unique" real-world entity that occurs in different contexts, as in a single customer who

transacts business in several different cities. For example, in a Customers dimension with levels

Country, State, City, and Customer, a customer named Acme Metals may transact business in

Chicago, Philadelphia, and Orlando. If so, a category named "Acme Metals" will exist under

each of these cities. As a result, the Customer level is not unique.

In order to join alternate drill-down paths, the category values in the convergence level must be

unique. Transformer is asking you to verify that is true for the convergence level that you are creating.



Flagging categories as unique by selecting the appropriate check box in Transformer does not

automatically make them so. You are merely specifying that this property is true for the source data

used in your model.

If you indicate that the categories in a level are unique when they are not, invalid results will occur,

because Transformer cannot reliably associate the non-unique values to the appropriate (convergence-

level) categories. When the source value is read, Transformer places the value in the first category

associated with that value. Duplicate categories in the level are renamed using the tilde character

(~), making them non-numeric.

It is possible that your source data might contain columns that populate levels you thought were

unique that are in fact not unique. In such cases, a message informs you of the potential problem

when you attempt to generate categories, but only if the data source contains all of the columns

required for the levels in question. If categories for some levels have values that are derived from

other transactional data sources, uniqueness conflicts may arise but be undetected.

If you change the optimization setting for the data source to Maximize Query Speed, the model is

not checked for uniqueness conflicts. (A message warns you that checking is disabled.)

Uniqueness and Missing Values

It is possible to declare that a level contains unique source values when it truly does, and still

encounter uniqueness problems. This arises when you create an "unbalanced" dimension in which

more than one category value in a level is missing.

For example, suppose you create a Regions dimension that contains the levels Country, State, and

City. It is reasonable to assume that the State level is unique. However, if some countries do not

have states, the State column for those countries will be blank and non-unique values will be

introduced (in the form of missing values). To avoid this kind of uniqueness violation, we recommend

that you create a subdimension for that portion of the dimension that does contain State values,

and designate that State level as unique. You can then omit the portion of the dimension with the

missing States.

Resolving Uniqueness Problems in the Source

If you are using Impromptu IQD files as your data source, you can use Impromptu to make the

categories unique. To do so, use Impromptu to create calculated columns that combine the original

non-unique source values with a qualifying code or identifier. You can use the resulting unique

column to provide unique category values to Transformer.

If you are using a data source other than IQD, you must use your IBM Cognos or third-party query

tool to render the level values unique, or you must refine the source data so that it does provide

unique values.

Measures Have No Values in Some DimensionsThere are two possible reasons for missing values in the PowerPlay clients:

● the source data contains missing values for the measure, in which case Transformer writes the

value 0 or the value "na" to the PowerCube.



● the measure in question applies to only a portion of a dimension, in which case Transformer

may not allocate the measure to lower levels.

To remedy the problem, you can:

● clean your source data so records with missing values are excluded, or so missing values are

eliminated.

● omit the measure from the PowerCube if there are many missing values, and if the measure is

not required.

● allocate the measure values to dimensions, or to lower levels and categories within a dimension.

Relative Time Calculations Aren't Evaluated CorrectlyThis can arise if you set up relative time categories but don't tell Transformer to include all date

categories. In order to calculate relative time periods correctly:

● All date categories must be included in the date dimension, even categories for which no source

transaction records exist. For each level in the date dimension, check that the Generate All

Categories in the Period check box on the Time tab (Level property sheet) is selected. Running

Generate Date Categories from the Run menu, in order to regenerate any missing date categories.

● The current period must be set correctly, and by the appropriate data source. For each data

source, check whether or not the Sets the Current Period check box on the General tab (Data

Source property sheet) is selected or cleared. Alternatively, set the current period manually.

● Date values in the levels in your date dimension must be sorted in ascending order. For each

level in the date dimension, ensure that the categories in the level are sorted in ascending order

based on the column used to populate the date dimension (use the Order By tab of the Level

property sheet).

● If you are expecting the calendar day to set the period of comparison, for example, if you want

YTD to contain all data up to 11/25/2002 and the prior YTD category to be based on the same

calendar day, then you must ensure that you have specified "day" as the lowest degree of detail

within the time tab of the column properties in the data sources.

Source Data Doesn't Match the Data Source in the ModelWhenever your source data changes (for instance, if you add new columns, or you delete or rename

existing columns) you can use the Modify Columns command to synchronize the model columns

with the new columns in your source data file.

Allocation Tab Is Disabled for All Objects in a ModelThe Allocation tab is not enabled unless your data sources are defined so that not all measures

apply to all portions of the dimension map. Allocation is possible only when

● your model uses multiple data sources, and

● the measure values from one or more of these data sources have meaning in some portion of

the dimension map only when Transformer allocates them.



Inconsistent Numeric Output After MigratingModels Created Using Native dBaseData Sources

When you open a model that was created in an earlier version of Transformer and that model uses

native dBase as the data source, unexpected numeric values may appear. This known migration

issue relates to the fact that, beginning in Transformer Version 7.4, dBase data sources must be

accessed through ODBC.

Steps to Avoid the Problem of Incorrect, Invalid, or Truncated Numeric Data

1. Review all columns in your native dBase models to identify which ones use numeric data types.

2. On the property sheet for each column that you identify, change the scale and decimal properties

as required to generate the correct output.

3. Save the model and confirm that the numeric data appears correctly in Transformer Version

7.4.

Tip: You can instead manually edit the older MDL file. For example, change the column definition

from Storage Default Scale 2 Size 12 Decimals 2 to read Storage Float64 Scale 0

Size 8 Decimals 0.

Transformer Seems to Ignore a Measure's Decimal ValuesFor datatypes other than Float, Transformer reads numbers based on the Input Scale defined for

them in the Column property sheet. If the Input Scale is set to 0 (the default), then Transformer

uses only the integer portions of column values. For example, if a source column contains values

such as 10.234 and 14.253, Transformer will by default read the values 10 and 14.

To include the decimal portion of an input value, you must scale the source values. You can do this

by specifying an Input Scale for the column associated with the measure on the General tab (Column

property sheet). For example, to scale input values like 10.234 and 14.253, specify an Input Scale

of 3. This causes Transformer to multiply the input values by 10 to the power of 3, yielding integer

values 10234 and 14253 that Transformer uses in the model.

To reverse the effect of the Input Scale so values are processed and displayed correctly, you can

specify both an Output Scale and Decimal Places for the measure. Using the previous example, you

could specify

● an Output Scale of 3, which causes Transformer to divide the integer values 10234 and 14253

by 10 raised to the power of 3, yielding 10.234 and 14.253, or

● a Decimal setting of 3, causing PowerPlay to display three decimal places.

The following table illustrates combinations of various Input Scale, Output Scale, and Decimal

settings for source data values, and the resulting display in PowerPlay.

Note: The storage type of the measures is 32 bit, but you can change this to 64-bit floating point

if you need to support a greater degree of precision. The overflow limit for this storage type is 18

significant digits, excluding the decimal separator. If your measure values have more significant

digits after the decimal place than is specified in the precision attribute of the model, the number

is truncated and the last bit rounds up, rather than down.



Tip: In situations where your source database supports a greater degree of precision, you can avoid

overflow errors in Transformer by multiplying your measures by 1. This calculation forces Trans-

former to apply the more restrictive limits associated with the 64-bit floating point storage type,

and to handle overflow errors in the expected manner.

What's displayedDecimalsOutput ScaleInput ScaleInput Value

10.23433310.234

10.2340053310.234

10.2323310.234

10.2423310.238

0.10033-3100599

0.113-3100599

75.292240.7529

Oracle Database Schema Definition Modifies Numeric Category LabelsIf your data source is an Oracle database, when you create category labels in Transformer based

on a numeric field, the numeric values will be scaled by two. For example, if your data source values

are 100, 100.23, and .14, when this column is used in the dimension as the label and categories are

generated, the category labels appear as 10000, 10023 and 14.

This is because Transformer creates category labels based on the database schema definition, and

the data source input scale cannot be overwritten by modifying the columns. As a result, all measures

defined in the cube will reflect the data source column input scale.

Unknown Signon Appears in Signons List with .iqd FileWhen you use an Impromptu Query Definition as a data source, Transformer uses information

within the IQD file to create a new Signon object within the model. The new Signon object is

assigned the logical database name that exists in the IQD file.

For example, if you use an .IQD file that looks like this:

COGNOS QUERY

STRUCTURE,1,1

DATABASE,OUTDOORS

DATASOURCENAME,C:\COGNOS\IMP30\WORKSPCE\PRODS.IMR

TITLE,PRODS.IMR

BEGIN SQL

select T1."PROD_LINE" as c1,

T1."PROD_TYPE" as c2,

T1."PROD_NO" as c3,



T1."COMMENTS" as c4

END SQL

COLUMN,0,Product Line

COLUMN,1,Product Type

COLUMN,2,Product No.

COLUMN,3,Product Note

Then Transformer adds a signon object named OUTDOORS to the Signons list. You can use this

signon to access the database automatically using the .IQD file if all of the following conditions are

met:

● you create a database object (in Series 7 Access Manager) with the same name as the Signon

in the currently configured authentication file

● you add a database signon for the database

● you associate that database signon with an Auto-access entry for a user that is currently assigned

to a user class, and

● you use the Authentication tab (Model Properties dialog box) to include Access Manager's

auto-accesses information.

If you are not using Access Manager and you want to embed the database user ID and password

(in an encrypted form) in the model, open the Signon property sheet for the new signon.

User Class Security Isn't WorkingAfter you have set up security by user class, you must associate the appropriate user classes to the

cube that you want to secure.

To do this, perform the following steps:

1. Drag the appropriate user classes from the User Class list to the appropriate cube in the

PowerCubes list.

2. Rebuild or update the cubes to which you have added user classes.

When you enable security by user class for a model, the unique internal identification numbers for

user classes in the active authentication file become embedded in the model. If you then define user

class views in Transformer, these views are based in the user classes associated with the unique user

class identification numbers.

You will lose any user class views you have established if either of the following conditions are met:

● You replace the local authentication export (.lae) file or namespace with a new one, even though

the new file is structurally identical to the original.

● You change the Access Manager Configuration (Authentication tab, Model Properties dialog

box) to point to a new authentication file.

As a result, you will have to reconfigure all the user class security you have defined in the model.

When making changes to user classes, make the changes to the original Authentication Configuration,

and then use the Update Access Manager Information command (Tools menu) to have Transformer

update its internal user class definitions based on the latest contents of the authentication source.



For example, suppose you are currently configured to use the authenticator file MYUSERS.AUT,

and you have set up several user classes within this document. If you then enable user classes within

Transformer and set up user class views, you cannot replace MYUSERS.AUT with another

authentication source (even one with the same name) and expect your existing security to work.

When or Why Should I Use Multiple Data Sources?There are two reasons for using multiple data sources:

● Performance. For high data volumes, you can read values for low-level categories from a data

source that contains only values for these lower levels and any associated measure values. By

doing so, Transformer directly associates column values with low-level categories in the model.

As a result, Transformer does not have to read and drill down through higher-level categories

that provide the structure for the model when creating PowerCubes.

Note: In order to associate column values directly with a level (without requiring that the

columns for ancestor levels be in the same data source), the level must be unique.

● Setting up measure domains and measure allocation. Measures with different domains require

different data sources. Since measures no longer have to apply to the entire dimension map,

you can use measures in new ways.

How to Design Your Data Sources

When setting up multiple data sources, it is advisable to create one data source for each dimension,

and then set up one or more data sources to provide measures for those dimensions. In this way,

you separate the data sources that provide the model structure (during category generation) from

the bulk of the data used to provide values for the model's measures (during PowerCube creation).

For example, consider the following hypothetical dimension map that contains 3 dimensions. If

you use multiple data sources, consider creating the following data sources:

● a single structural data source for each of Dimension1, Dimension2, and Dimension3

Note: Simple dimensions, and most date dimensions, do not require their own data sources.

● one or more data sources with measure values referencing leaf levels (or some summary level)

in each dimension

The relation of the data sources to the dimension map is shown in the following diagram:

Each structural data source provides category values for the levels in a single dimension. Each data

source provides category values for the lower levels (not necessarily the leaves) in each dimension,

plus the model measures.



Structural data sources do not have to provide level columns down to the lowest level in the

dimension. In the preceding example, it would be possible for Data Source 3 to include columns

only down to level 3.3. In such a case, the data sources themselves can provide category values for

levels 3.3, 3.4, and 3.5.

ObjectStore VirtualAlloc Failure Prevents Startup of TransformerIf there is insufficient storage space for the ObjectStore, you will see the following error message

when you try to start Series 7 Transformer on Windows XP SP1 or Windows 2003:

ObjectStore internal error.

VirtualAlloc failed to allocate at any address - not enough storage is available to process this com-

mand (8)

The problem does not seem to arise with Impromptu Web Reports, although it also uses ObjectStore.

If you experience the problem, we recommend that you do the following:

1. From the Control panel, click System and, on the Advanced tab, click Environment Variables.

2. Ensure that OS_AS_START is not defined among the environment variables. That allows the

system to determine an optimal PSR (Persistent Storage Region) for the client address space.

The default range is from 0x30000000 to 0x38000000.

3. If there is not already a system environment variable called OS_AS_SIZE, create one.

4. Test various settings for the OS_AS_SIZE environment variable, as the required setting depends

on the amount of memory on your computer, your operating system, and your components.

● With Transformer 7.3 or 7.4 on Windows XP SP1 or Windows 2003, try setting the value

at 0x08000000 (128 MB).

Note: To translate the hexadecimal value into its equivalent in megabytes, cut 5 trailing zeroes

to divide by 0x100000 (which represents a megabyte). The result is the number of megabytes

in hexadecimal notation. For example, 0x20000000 divided by 0x100000 equals 0x200, which

in decimal notation represents (2x16x16) =512 MB.

5. Lower the value of the OS_AS_SIZE environment variable in small increments, rebooting your

computer and re-applying your new configuration after each test.

6. Repeat as necessary, until the problem no longer arises.

Allocation Appears Incorrect in Time-based Partitioned Cubes if a Cube is Updatedwith Data from Another Period

After you update an individual or member cube with data from another period, the allocated

measures may appear incorrect because Transformer allocates measures at the lower levels of a

dimension and then rolls them up. By design, measures are not separately allocated at each level.



Unable to Import IBM Cognos 8 Data that Contains TEXTBLOB, CLOB, or BLOBData Types

Transformer cannot import an IQD data source from Framework Manager if it contains TEXTBLOB,

CLOB, or BLOB data types. Before externalizing IQD columns or query subjects from Framework

Manager, ensure that they do not contain these data types.

Insufficient Cache Space for Proper Cube ProcessingIn some cases, the default cache size may not be sufficient to support cube processing in your par-

ticular Transformer environment. If you suspect that Transformer is running out of space, you can

recalculate your cache size requirements and adjust the default setting accordingly.

The amount of space needed for the disk cache depends on the number of records in your source

data, the number of categories in your model, the operating system you are running, and the amount

of RAM in your computer. However, you can use the following formula as a guideline for

determining the size of your disk cache:

Disk Cache = APM - (Model + CGen) - C

where APM is the available physical memory on the computer; Model is the size of the .py? file;

CGen is the amount of memory allocated to cube generation using WriteCacheSize (an .ini file set-

ting); and C is the amount of memory used by the operating system, executables, and so on. All

measurements are in megabytes.

The model occupies physical memory at different stages during cube creation. These stages include

source data processing and metadata updating. During the data update stage, the virtual memory

manager pages the model out of memory and should have a minimal effect on the size of the disk

cache.

Adjusting the cache sizes may significantly alter performance in Transformer, the Series 7 PowerPlay

clients, or the IBM Cognos 8 Web studios. If the results are not satisfactory, restore all settings to

their default values.

In Series 7, these are controlled by changing the values of

● a Transformer .ini file setting (Windows)

● read/write memory settings in server Transformer (UNIX)

● cube-specific READ_CACHE settings in PowerPlay Enterprise Server (Administration tool)

● write_cache, read_cache, and object_cache settings controlled by Configuration Manager

In IBM Cognos 8, the server-wide READ_CACHE default applies. For more information, see your

operating system documentation.

Unable to Connect to the DatabaseWhen you try to execute reports or queries immediately after installing and setting up a connection

to DB2, Sybase Adaptive Server Enterprise, or Teradata, you may receive an error message that

there is a problem connecting to the database or executing the report.

You must reboot your system to solve this problem.



With Transformer on UNIXTransformer for UNIX (that is, the rsserver executable) does not support the full range of features

carried by Series 7 Transformer (Windows), including client-server mode. For more information,

see the "Features and Editions" topic in the Welcome chapter.

Client-Server Connection Is Not WorkingTransformer communicates with Transformer on UNIX via the PowerGrid network daemon (netd)

on the server. When sending information to and receiving information from PowerGrid, it is possible

to encounter communication problems.

There are several things you can verify to ensure that Transformer can communicate with netd on

the server, and that netd can start the rsserver program.

● Check that there is an active netd process running on the server. For example, you might enter

a UNIX command like

ps -ef | grep netd | grep cognos

Note: The ps command varies from one UNIX system to another. Use the equivalent for your

system.

● Check that the service port number defined for the PowerGrid network daemon on the server

matches the one you have used in your connection definition. The environment variable on the

server is NPNETD, and it is set to a value like ISnnnn. The four-digit nnnn value must match

the setting on your PC.

● Check that the following environment variables are set, to have your UNIX server continue

processing, even if your Transformer (client) users have closed their PowerGrid session:

ServerWaitPeriods=0(the default is 30 seconds)

ServerAnimateTimeOut=1 (the default is 3 seconds)

Can't Create a Model Using Transformer on UNIXIf you are unable to create a model on the server using Transformer on UNIX, check that

● You can run rsserver from the command line on the server.

● A connection exists and is working between Transformer on Windows and Transformer on

UNIX.

● Transformer on UNIX is configured to create models and PowerCubes in locations to which

you have write access.

● The Connect.ini file (derived from Cognos.ini) contains the following entry:

[Service - Transformer Server]

NETWORK=rsserver.sh

This entry is used to locate a server shell script that PowerGrid uses to launch Transformer on

UNIX. Ensure your server account can locate and execute this file.



With PowerPlay Client ApplicationsThe following notes pertain to Series 7 operations. For information about using Transformer

PowerCubes in IBM Cognos 8, see the Framework Manager and Analysis Studio documentation.

Production Reports No Longer WorkCertain changes you make to a Transformer model will affect, and possibly disable, production

reports derived from PowerCubes based on that model.

The following changes will require you to recreate production reports using a new version of the

cube (.mdc) file:

● If you add or remove a dimension from the model, any production reports based on that model's

PowerCubes will no longer function. You will have to recreate the reports using the revised

cube.

● If you remove or rename categories used in a production report, then values for those categories

will be displayed as the value "error" in PowerPlay. You must manually add any new categories

to the report.

● If you replace the Access Manager configuration currently being used by the model, then any

existing security by user class will no longer apply. You will have to redefine your user class

views and recreate all PowerCubes.

Dates and Numeric Values Not Formatted as ExpectedData formats depend on the locale configuration settings for your IBM Cognos Series 7 product.

By default, IBM Cognos Series 7 server products and IBM Cognos Impromptu use a data format

configuration file named cerlocale.xml. IBM Cognos Visualizer, IBM Cognos PowerPlay for Win-

dows, and IBM Cognos Transformer use Windows Regional Settings for locale settings and data

format information. However, these products can also be configured to use the data format config-

uration file.

For information about managing data formats, copying data format information, or editing the

data format configuration file, see the Configuration Manager User Guide, or contact your

administrator.

Inadequate Query Performance with Partitioning in PlaceIf you have followed the documented partitioning guidelines and query performance is still poor,

collect the following information:

● the Transformer log file from the cube build; shows how dimensions are partitioned and how

well records are being consolidated

● the populated model (.mdl or .py? file) and cube (.mdc file); the partition status shows partitioned

categories and record distribution

● an indication of which dimensions are most often queried, and the extent to which users of the

PowerPlay client applications are querying at lower levels in your cube



● details about problematic reports or exploration activities, such as slow performance when

navigating alternate drills or nested levels in the PowerPlay client applications

We recommend that you forward this information to Cognos Software Services and let our experi-

enced personnel investigate the problem.

PowerCube Access from PowerPlay Clients Is Too SlowIf your users of the PowerPlay clients are finding that response time is unacceptable because the

PowerCubes they are accessing are very large, consider the following measures:

● Ensure that the users of the PowerPlay clients configure their PCs correctly. In particular, ensure

that their Cognos.ini files contain sufficient entries for the PowerPlay Data Server cache size.

● Partition the PowerCube so PowerPlay spends less time computing summary values.

● Wherever possible, consolidate the data in the PowerCube to reduce its size and improve

response time.

Summaries Don't Match Individual Cell ContentsProblem: In PowerPlay, summary row and column values do not match the contents of the individual

cells.

The way you define measures in Transformer affects the way in which they are rolled up in the

PowerPlay clients.

For example, if you have allocated measures proportionally (based on other measures) or as constant

values, then the values displayed in the PowerPlay clients do not add up to (or average to, in the

case of average rollups) the summary values for rows, columns, or layers. If this is a concern for

the people who will use the PowerCubes, consider using the Description tab to add explanations

for the measures affected. The PowerPlay client users can use the Explain feature to read explanations

of why the rollups are not working as they might expect.

Cell values always coincide with summary row and column values based on the rollups being

applied. For averages and weighted averages, the averages of the summary row and column will

not be the average of the entries in each cell. Instead, it will be the average based on all of the records

that contribute to the averages in each cell.



Appendix F: Error Messages

On the Index tab (Help Topics dialog box), type or scroll to the corresponding message number.

Note: All Transformer message numbers begin with "TR".

Message TR0104A user was expected but not found. Either you have referred to a user by an object nameor object identifier that Transformer is unable to locate, or you have not referenced theuser.

This error occurs when Transformer's Model Definition Language (MDL) requires an object and

is unable to find it. In this case the object is a user class. The problem may be that the user class

was not referenced, or that it was referenced with an incorrect object name or object identifier.

For details about MDL syntax, see the Transformer MDL Reference.

You can verify object names and object identifiers in the user interface. To make them visible, select

the Object Name and Object Identifier check boxes in the Preferences dialog box (Titles tab). The

object name and identifier will appear in the title bar at the top of each property sheet, and will

also appear in a ToolTip when your cursor hovers over an object.

Message TR0105The view of an ancestor user class restricts the operation you requested.

You have attempted to override a security setting that was imposed at a higher level in the user

class hierarchy. Security for a user class is inherited from its ancestor user classes.

If you require access to a specific category that is inaccessible due to ancestor user class views, you

can

● remove the restrictions imposed by the ancestor user class

● create a new user class and appropriate user class views at the ancestor level

● selectively add users from the lower-level user class to the ancestor user class so that these users

can view the categories that are accessible by that class

Message TR0106User class views are not allowed to perform the operation you requested.

You have attempted to perform an operation in a user class view that Transformer does not support.

See the Transformer online Help for more information about operations involving user classes.


Message TR0107You cannot view the categories in a dimension that an ancestor user class has restricted.

This message is most likely to appear in the log file when you make updates to a model using an

MDL script. Under normal circumstances, the Transformer user interface prevents you from viewing

all of a dimension for which a custom user class view is applied in an ancestor user class. Verify

that the MDL script is correct, and that you have specified the correct object identifier for each of

the objects referenced.

If the user class for which you want to view the entire dimension does require access to that

dimension, you can

● Change the properties of the user class view for ancestor user classes so that all categories are

included.

Note: All members of the subordinate user class (the one for which you are trying to include

the measure) will gain access to the entire dimension.

● Add specific users from the subordinate user class (the one for which you are trying to include

the measure) to the ancestor user class in which the user class view is defined. This will give

these users access to all information that is accessible by the ancestor user class.

● Create a new user class for which the categories are included, and add users to it as required

to meet your needs.

Message TR0108An ancestor user class excludes one of the measures you include.

This message is most likely to appear in the log file when you make updates to a model using an

MDL script. Under normal circumstances, the Transformer user interface prevents you from

including a measure that is excluded in an ancestor user class. Verify that the MDL script is correct,

and that you have specified the correct unique ID number for each of the objects referenced.

If the user class for which you want to include the measure does require access to that measure,

you can

● change the properties of the ancestor user class in which the measure is excluded.

Note: All members of the subordinate user class (the one for which you are trying to include

the measure) will gain access to the measure when it is included.

● add specific users from the subordinate user class (the one for which you are trying to include

the measure) to the ancestor user class in which the measure is included. However, this will

give these users access to all information that is accessible by the ancestor user class.

● create a new user class for which the measure is included, and add users to it as required to

meet your needs.

For more information, see the Access Manager online Help.



Message TR0109Transformer couldn't read the data source.

Ensure that your source data file is valid, and that it is in the appropriate format for the type of

data source defined within the model.

If you are using an Impromptu query definition file (.iqd) as a data source with a 32-bit ODBC

driver, this error may occur when the driver setup contains a transaction level of Repeatable Read

and no security information. Try changing the transaction level to Read Uncommitted; or enter a

user id and password in the Client Access driver setup.

If you get this message when trying to access Transformer from Impromptu, and also receive the

Oracle error message ORA-3126, the problem may be that the SQL*Net connect string to Impromptu

in tnsnames.ora is incorrect. You could also be using an incompatible version of SQL*Net.

Message TR0110No records were found in data source < >, the file is empty.

You have attempted to generate categories from an empty input file. The file exists, but contains

no records that Transformer can use. It is likely that one of your source data files has been inadvert-

ently overwritten by an empty file, or that the process that creates your source data files has failed

and created a file with no records.

Check that the source data file associated with your model's data sources contain the data they

should. You can find the name of the file associated with a model data source in the Data Source

property sheet (Source tab).

Message TR0111< > is not a valid file name.

You have specified an invalid filename for a PowerCube or a source data file that is related to a

data source.

Check the name of the file you entered, and verify that it is correct.

Message TR0112There isn't enough memory available.

There is insufficient memory for Transformer to perform some operation.

Windows Applications

Try closing as many open applications as possible. If this fails to correct the problem, save your

model, exit, and then restart Transformer. You may need to restart Windows.

If this is a recurring problem, you may need to add memory to your PC in order to continue

working on the model or PowerCube.



UNIX Server Applications

Try lowering MaxTransactionNum. For more information, see MaxTransactionNum in the

Transformer for UNIX Guide.

On an HP9000, the problem might be the data region setting. Try increasing the data region by

increasing the MAXDSIZ HP9000 kernel setting. For more information, see your UNIX adminis-

trator or manual. You may also need to increase the PPDS memory setting. For more information,

see the Transformer for UNIX Guide.

Message TR0113Transformer can't create the file < >.

Transformer is unable to create a file. Ensure that you are not attempting to overwrite a read-only

file, and that the file is not currently locked by another process.

Message TR0114Transformer can't write in the model temporary file.

This error is usually caused by insufficient disk space to write the temporary work files. For a rough

estimate of how much disk space you need to create a cube, multiply the size of the data source(s)

by 3.5.

You can find the location of temporary work files in the Directories tab of the Preferences dialog

box (File menu). Free up more space on the work file drive and retry.

For more information, see the Administrator's Guide.

Message TR0116Transformer can't open the file < >.

Make sure that the path and file name are correct.

There may be incorrect information in the Directories tab of the Preferences dialog box (File menu).

If this error occurs when you are running an MDL script in the command line, the problem may

be that the MDL script contains the command SavePY and you are also using the -s command line

switch to save the file to .py? format. Try removing the -s option.

Message TR0117Transformer can't open the file < >.

You have attempted to generate categories or create a PowerCube, and Transformer is unable to

find one of the source data files associated with a model data source.



In the Source tab (Data Source property sheet), verify that the Local Data File specified for each

data source exists, and that you have access to the drives and directories in which these files are

stored.

If you have specified an alternate source file for the PowerCube in the PowerCube property sheet

(General tab), ensure that the alternate (transactional) source file exists.

If you are building cubes in a multi-cube model, ensure that the source file information is complete

and correct for all cubes, including disabled cubes.

Message TR0118Transformer can't read the database < > defined in < >.

Transformer can't open the data base that is specified for a data source.

If you are using an Impromptu query definition file (.iqd) as the data source in a Transformer model,

● ensure that the query does not exceed the SQL limit. Try removing some of the columns in the

Impromptu report that the IQD is based on.

● ensure that the IQD file was created as Local IQD.

● confirm that appropriate access was granted to the Impromptu query definition file (.iqd) file.

● confirm that the Impromptu report that created the Impromptu query definition file (.iqd) can

run successfully in Impromptu.

● ensure that the qualified database table names in Impromptu are 30 characters in length or

less.

● turn off Cache Query Results in Impromptu.

● ensure that the Impromptu query definition file (.iqd) was not created against hotfile tables in

Impromptu. Hotfiles are not supported by Transformer.

● clear the Minimize Connect Time to the Database check box in the Client-Server tab of User

Profiles (Catalog menu) in Impromptu.

● make sure that the database user ID and password are valid.

If you are attempting to use an Impromptu query definition file (.iqd) as a data source for Trans-

former and it works on the client but not on the server, open the Impromptu query definition file

(.iqd) in a text editor and check the syntax and the data source name. If the file was created in a

previous version of Impromptu, the report that generated the IQD must be rerun in the latest version

of Impromptu that is installed on the system being used.

If you are attempting to generate a PowerCube on an HP-UX 10 server computer using rsserver,

and the data source is an Impromptu query definition file (.iqd) that accesses a Sybase database,

ensure that the following requirements are met:

● Open Client must be installed on the HP-UX server computer.

● The Sybase and DSQuery environment variables must be set.



● The Sybase /bin directory must be in the path.

The entry in the SQL.ini file that is referenced on the client must also be present in the interfaces

file on the UNIX machine.

If you are creating a cube in client-server mode, Sybase and DSQuery must be set in <install_dir>/

powergrid/tcpip/netbin/rsserver.sh before the setpya.sh script is referenced.

If you are using the IBM Client Access ODBC driver, turn off Extended Dynamic Support. In the

Client Access ODBC configuration utility, change *CHG to *None.

If you are attempting to generate server PowerCubes from the client using Oracle data sources,

● ensure that the connect string in the tnsnames.ora file on the client computer matches both the

connect string used in the logical database definition in Impromptu and the connect string in

the tnsnames.ora file on the server computer where Transformer is running.

● ensure that the required Oracle environment variables are set in rsserver.sh. For more informa-

tion, see the Transformer for UNIX Guide.

Confirm that the Databases section of the Cognos.ini or COGCONNECT.ini file has a database

connection string to the relevant database.

Message TR0128Only data source input files of the following types are supported in this mode<data_source_list>.

You are using the server version of Transformer and specifying that processing is to occur on the

server using a server data source. In order for this to happen, you must provide a server-based data

source.

Message TR0131The argument for command line option -k is invalid.

When trying to pass database login information to Transformer using the -k program parameter,

an invalid entry has been detected.

Ensure that you enter the value for the -k option properly.

Message TR0132The database name in the command line is blank.

When trying to pass database login information to Transformer using the -k program parameter,

an invalid entry has been detected.

Ensure that you enter the value for the -k option properly.



Message TR0133Unable to delete file < >.

Transformer is unable to delete a model file. Ensure that the file is not currently locked by another

process, and that it is not a read-only file.

Message TR0137Cube Group < > (and possibly others) contains cubes based upon key orphanage cat-egories. Due to the temporary nature of these categories, the data in these cubes maybelong to other cubes within their respective cube groups.

Transformer has created a key orphanage in the target level for a cube group. This can happen in

models for which the following conditions exist:

● the model uses multiple data sources

● the model contains one or more cube groups

● the data in the model's data sources does not provide sufficient information to connect a category

to one of the target categories in the cube group.

In the PowerCubes list, Transformer will generate a cube definition for the key orphanage category

within the cube group. You cannot delete this cube. However, you can add additional information

to the model that will eliminate both the key orphanage and the orphaned cube during cube

regeneration.

In some cases, it is possible that Transformer will automatically fix the problem. If subsequently

processed data sources contain data that can provide the path from the orphaned category to a

source category in the target level, then the potential problem will be fixed.

If the problem is not fixed automatically, to remedy the problem, you can:

● manually assign the categories in the orphaned category to one of the others in the target level,

provided you know which one they belong to, or

● determine what data is missing in your data source. You can then repair the original data source

or add a new data source with additional information that will complete the path from a category

in the target level to the orphaned categories.

If a key orphanage is created and subsequently deleted during a single cube generation process, the

resulting cube group will have an invalid status. Transformer can eliminate a key orphanage when

all of its descendants are placed under other categories in other cubes within the cube group. In

such cases, you must regenerate the cube group so that the data is placed back into the correct

cubes. For more information, see "Message TR0713" (p. 325).



Message TR0149A data input conversion or overflow error has occurred.

This error occurs when the data for a measure is too large for the storage type that is specified for

it. This problem often occurs when you have a measure that includes decimals, or if you use a large

input scale or output scale.

To solve this problem, first attempt to reduce the size of the Input Scale property on the Column

property sheet (General tab) and the Output Scale property on the Measures property sheet (General

tab). If this doesn't work for your model, change the Storage Type property on the Measures

property sheet (General tab) to a larger storage type.

Message TR0151One '< >' association from '< >' data source is already defined.

A second '< >' association from the same data source is not allowed.

This error occurs when you try to specify a second association of the same type from the same data

source to an object.

To solve this problem, either remove the association that you don't want, or make an association

to another data source.

Message TR0202You didn't include any dimensions. Transformer can't create a PowerCube.

You have attempted to create a PowerCube for which all of the dimensions are omitted.

Open the PowerCube property sheet and click the Dimensions tab. Ensure that not all dimensions

are set to Omit Dimension.

Message TR0203You didn't include any measures. Transformer can't create a PowerCube.

You have attempted to create a PowerCube for which all of the measures are omitted.

Open the PowerCube property sheet and click the Measures tab. Ensure that not all measures are

set to Exclude.



Message TR0205Double quotes were changed to single quotes in the source value, label, description, sortvalue or other text field. This avoids syntax errors in the MDL Language.

When Transformer generates MDL, it uses double quotes to define the model. This error message

indicates that double quotes in the source data have been changed to single quotes so that Trans-

former can distinguish between the source data and the structure of the model.

For more information, see the Transformer MDL Reference.

Message TR0206You didn't include any dimensions in PowerCube < >. The cube wasn't created.

You have attempted to create a PowerCube for which all dimensions are omitted.

In the PowerCube property sheet (Dimensions tab), ensure that at least one dimension has a setting

other than Omit Dimension.

Message TR0207Data source < > is not related to any dimension so it cannot be processed.

You have attempted to generate categories using a data source that is not associated with any level

in any dimension in the model. This can happen when a data source has been added to the model

but no dimension has yet been set up for that data source.

Message TR0208Consolidation for the PowerCube <> (and <n> others) is suppressed because a before-rollup calculation was also defined.

Whenever a model includes a calculated measure with a timing of Before Rollup, consolidation

must be suppressed for cubes in which that measure is included. Consolidation uses specified rollup

settings to write consolidated records to a PowerCube. If consolidation were performed, the level

of detail in the PowerCube's records would not contain sufficient detail to perform the calculation

before the rollup is applied.

If you want consolidation to be performed, ensure that the PowerCube does not include any calcu-

lated measures with a timing of Before Rollup.

Message TR0209Sorting and consolidation for the PowerCube <> (and <n> others) must be performedto determine correct state measure values.

The settings in the model are such that consolidation is required. Consolidation is required, in spite

of any settings for consolidation in the PowerCube property sheet, under the following conditions:



● a setting other than Default is specified for Duplicates rollup, and

● consolidation is set to anything other than pre-sorted.

Message TR0210<n> data input conversion or overflow errors occurred. See the log file for details.

Within one of the source data files that is associated with a model data source, the value specified

for a measure is either too large or uses an invalid storage type.

If you check the log file associated with the model you are updating, you will find a line that looks

something like this:

(TR1703) A data input conversion or overflow error occurred at source record number 100 for

measure 'Revenue' in source file 'installation_location\national.asc'.

You can use this information to identify which input data records contain invalid source values.

Message TR0214The temporary file for source file < > and PowerCube < > is empty. Check that yoursource file contains data.

During the PowerCube creation process, Transformer has been unable to find data for use in a

PowerCube. This can happen if

● your source data file is empty.

● you have applied a view to the PowerCube that excludes all of the data for a given dimension.

Excluding all data for a dimension using a dimension view will cause the PowerCube to be

empty.

● the timing for all data sources is such that none of them are used to create PowerCubes.

● the data source containing measures for your model is associated with categories in an alternate

drill down. For more information, see "Message TR1320" (p. 337).

● there are BLOB fields in the source file. Transformer does not support BLOB fields.

If your source file is an Impromptu query definition file (.iqd), try running the report in Impromptu

to make sure that data is being returned.

Message TR0215The PowerCube < > (and <n> others) may require consolidation. Consolidation can'tbe performed because of before-rollup calculations or because a cube is designated asusing Direct Create.

There are conflicting settings in the model that simultaneously require consolidation to be performed

and prevent it from occurring. This happens when calculated measures exist with a Rollup Timing

of Before Rollup.



The reason that these settings conflict is that the consolidation will eliminate from the PowerCube

the records required to perform the calculation before rollups are applied to them.

In this case, Transformer does not perform consolidation.

Message TR0217Data source < > is turned off for cube generation. The process has aborted.

The PowerCube Creation box on the Data Source property sheet (General tab) is not selected. This

occurs with structural data sources because they do not contain transactional data.

Message TR0301There are no columns for AutoDesign to use.

Transformer is unable to locate columns for use in the AutoDesign process.

Check that the columns in your model's data sources have not been inadvertently deleted.

Message TR0303Transformer has detected invalid allocation specifications. These allocations will beremoved. Do you want to continue?

A calculated measure was allocated and then deleted from the model. After you choose Yes to

continue, Transformer will change the allocation on which the measure was based to N/A (if it is

to a level), or Constant (if it is to a dimension).

Message TR0404You didn't specify a server data source. Do you want to continue?

This warning message appears only for Transformer on UNIX.

If you attempt to generate server categories, Transformer issues this message if you have not specified

a data source on the server that Transformer on UNIX can use to obtain source values.

To resolve this problem, specify a data source on the server computer.

Message TR0407The model you imported has level or category Inclusion set to Remove When Unused.This option is no longer supported. The Inclusion property has been set to the default,Always Include.

Some options that were supported in previous versions of Transformer version are no longer sup-

ported.

Verify that the Inclusion property for all levels and categories is set correctly.



Message TR0408This model was built with a system language setting of <> that differs from the currentsetting. The language setting dictates the sorting rules used in category ordering. Doyou want to resort all ordered categories?

Your PowerCube will be generated successfully despite this error.

A model populated on a computer with one language setting was moved to another machine with

another language setting.

Message TR0412Error encountered while trying to save preferences to <path\trnsfrmr.ini>.

The most likely cause of this error message is that you have insufficient space on your hard disk.

Creating more space or switching to another drive may solve the problem.

Other possible causes are that the Trnsfrmr.ini file has read-only access specified, or the file is open.

Message TR0420A PowerCube group was expected but not found. Either you have referred to a PowerCubegroup by an object name or object identifier that Transformer is unable to locate, oryou have not referenced the PowerCube group.

This error occurs when MDL requires an object and is unable to find it. In this case the object is a

PowerCube group. The problem may be that the PowerCube group was not referenced, or that it

was referenced with an incorrect object name or object identifier.






Message TR0423A calculation definition was expected but not found. Either you have referred to a cal-culation definition by an object name or object identifier that Transformer is unable tolocate, or you have not referenced the calculation definition.


calculation definition. The problem may be that the calculation definition was not referenced, or

that it was referenced with an incorrect object name or object identifier.








Message TR0476A category set was expected but not found. Either you have referred to a category setby an object name or object identifier that Transformer is unable to locate, or you havenot referenced the category set.


category set. The problem may be that the category set was not referenced, or that it was referenced

with an incorrect object name or object identifier.






Message TR0500You can't move a category outside its drill-down path.

You have attempted to move a category from one drill-down path to another in the dimension

diagram for a dimension that contains an alternate drill-down structure.

You can move categories only within the scope of their own drill-down paths.

Message TR0501A regular category can't be the parent of a special category.

Under normal circumstances, the Transformer user interface does not allow you to make a regular

category the parent of a special category. This message normally occurs as a result of an action

applied by an MDL script.

Verify that the ID numbers you have used in the MDL script refer to objects for which only valid

actions apply.

Message TR0502You must specify a category code when you create a manual category.

You have attempted to create a category manually in a manual level and you have omitted the

category code.

Transformer automatically assigns a unique category codes to each of the categories you create.

The category code is displayed in the Category Code box of the General tab (Category property

sheet). Do not delete the category code.



Message TR0503Another category already has this category code. Each category requires a categorycode which is unique within the dimension.

When creating or modifying a category manually in a manual level, you have changed the category

code so that it conflicts with an existing category code. Transformer automatically assigns category

codes to each of the categories you create. If you modify this code, ensure that it does not conflict

with an existing category code.

The category code is displayed in the Category Code box of the General tab (Category property

sheet). Do not delete the category code.

Message TR0504You must specify a source value for a category in a source level. You can do this fromthe Source tab in the Level property sheet.

You may have inadvertently deleted the source value associated with a category. Source values are

required for categories in all source levels. Categories in manual levels derive their source values

from their category codes.

To solve the problem, you can edit the Source Value in the Category property sheet to provide the

correct value, or you can delete the category and allow Transformer to regenerate it.

Message TR0505This action would result in two categories with the source value < > under differentparent categories in level <level_name> which is designated as unique. The action iscancelled.

You have attempted to specify that the categories within a level are unique when Transformer has

verified that they are not. This can occur when you

● create an alternate drill-down and allow Transformer to designate the convergence level as

unique

● choose the Unique check box in the Source tab of the level's property sheet

In order for a level to connect multiple drill-down paths in an alternate drill-down structure, the

category values in that level must be unique. Similarly, in order for Transformer to be able to directly

relate column values to categories in a level, the category values in that level must be unique.

If you require that the level be a convergence level, or if you want to create a thin data source that

contains a column for the level without the ancestor levels, you must make each category in that

level identifiable by its value alone.

For example, if the convergence level is City in a Regions dimension that contains levels Country,

State, and City, ensure that in your source data file you somehow qualify identically named cities.

If the city Burlington occurs in the states Massachusetts and Vermont, you should add a qualifier

to the city name in the source data, as in Burlington (MA) and Burlington (VT). This may require



you to modify the values in source data files associated with one or more of your model's data

sources.

This error message can also occur if you change the source value of a date category, resulting in

two identical date categories in a level.

Message TR0507The share reference <category> isn't a category in the same dimension.

You have attempted to apply a share object using a level or category that is in a dimension other

than the one in which the share object is being applied.

Ensure that when you set the share object, you set it to a level or category in the same dimension

as the categories to which you'll apply the share object.

Message TR0508You can't move a category into or out of a convergence level.

On a diagram for a dimension with multiple drill-down paths, you have attempted to move a category

either out of or into a convergence level.

The categories in a convergence level cannot be moved out of that level, and no new categories

(from other levels) can be moved into the convergence level.

Message TR0510The share category ID is invalid. A share category ID must be the object identifier ofanother category in the same dimension.

You have attempted to specify an invalid share object. The share object

● must be an existing category

● must be at a higher level in the same dimension

● must be specified by its object identifier

To view the object identifier for each object, select the Object Identifier check box in the Titles tab

of the Preferences dialog box. The object identifier will appear in the title bar at the top of property

sheet, and will also appear in a tooltip when your cursor hovers over an object. You can also find

object identifiers in the .mdl file.

Message TR0514You can't delete or move the root category of a dimension.

The root category provides much of the information that is required by the dimension, including

the label that is provided for display in PowerPlay. You cannot delete a root category.



If you want to delete the entire dimension, choose the dimension on the dimension map and then

choose Delete Item from the Edit menu.

Message TR0515In a regular diagram, you can only connect to categories in a level above or below theoriginal level. You cannot connect to categories in the same level.

You have attempted to connect two categories in the same level to one another.

You can only connect a category to an ancestor category or to a descendant category. Verify that

the category to which you want to connect is in the level you want it in. If not, you'll have to

restructure your dimension so that the levels are positioned in the dimension hierarchy as you want

them to be.

Message TR0518You can only create subdimensions below categories that are in a source level.

You have attempted to create a subdimension based on a category in a manual level. Transformer

does not permit the creation of subdimensions below a manually-created category.

To create a subdimension, choose a source category.

Message TR0519A subdimension can't be rooted in this level. This level is above a convergence level ina dimension that contains alternate drill-down paths.

You have attempted to create a subdimension within one of the drill-down paths in a alternate drill-

down structure. Subdimensions are permitted

● in dimensions with only one drill-down path

● only at or below a convergence level in a dimension that contains alternate drill-down paths

Message TR0523You can't move this category because its position is determined by its order value prop-erty.

You have tried to move a category manually on a diagram, and the category's Order By value has

been set by one of the following methods:

● an Order By column was specified for the level in which the category resides

● an explicit Order By value was specified for the category in the Order Value box on the category

property sheet.

You can only move categories for which no Order By value is specified.



To change the order of categories for which there is an Order By column specified, do one of the

following:

● Choose another Order By column for the category's level. Transformer will re-order the cat-

egories based on the new Order By column.

● Disable the current order by settings by choosing the blank entry at the bottom of the Sort-By

Column list (on the Order By tab of the Level property sheet).You can then order the categories

manually by moving them on the diagram. You can then re-enable the order by settings.

Message TR0524Rollup can only be disabled for special categories.

Under normal circumstances, the Transformer user interface does not allow you to disable category

rollup for a regular category (the Category Rollup check box is only displayed for special categories).

This message normally occurs as a result of an action applied by an MDL script.

Verify that the ID numbers you have used in the MDL script refer to objects for which only valid

actions apply.

Message TR0525A category was expected but not found. Either you have referred to a category by anobject name or object identifier that Transformer is unable to locate, or you have notreferenced the category.


category. The problem may be that the category was not referenced, or that it was referenced with

an incorrect object name or object identifier.






The problem may be that you have not used the correct MDL object name for the category. Category

names are not the same in MDL and the user interface. In the user interface, category names are

derived from category labels. In MDL, category names are derived from category codes. Both the

category code and the category label appear on the category property sheet. For more information,

see the Transformer MDL Reference.



Message TR0528The source value for a date category in a month level must be numeric. Also, it must bein the format MMDD, unless specified otherwise in the Date tab of the column propertysheet.

You may have inadvertently changed or deleted the source value for a date category. Check the

Source Value on the category property sheet (General tab). Ensure that it is in the date format spe-

cified on the Time tab of the column property sheet.

Message TR0534This drill category cannot be deleted in a dimension with a single alternate drill-down.

You have attempted to delete the last drill category in a dimension. Each dimension must contain

at least one drill-down path. You cannot delete all drill-down paths.

To delete a drill-down that is the only one in a dimension, you can either:

● delete the entire dimension and, if desired, recreate it with the new drill down structure and

levels you want

● create an alternate drill-down structure, and then delete the drill category you no longer want

Message TR0535A category can't be created below a convergence level when the parent of that categoryis above the convergence level.

You have attempted to create a new category below the convergence level of an alternate drill-down

structure, and the new category descends from a category above the convergence level of an

alternate drill-down structure. Transformer does not allow this.

You can create new categories that descend from categories in or below the convergence level. You

can also create a new category in the convergence level by dragging from an ancestor level into the

convergence level.

Ensure that you have not dragged the mouse pointer too far to the right if you are trying to create

a new category in the convergence level.

Message TR0536Partitioning can only be specified for categories in a primary drill-down.

You have attempted to specify partition numbers for categories in the alternate drill-down path of

a dimension with an alternate drill-down structure. Transformer supports partitioning within the

primary drill-down only.

To eliminate the problem, isolate which of the categories in the alternate drill-down path have

partition numbers specified, and then change their partition numbers to 0.



Tip: If you are unsure which of the categories in the alternate-drill-down path are partitioned, an

easy way to isolate categories with partition numbers other than 0 is to save the model in MDL

format. You can then open the MDL file using the text editor of your choice, and search for the

string NewPartition. Any levels or categories with partition level numbers other than 0 will show

up as, for example, NewPartition 1.

Message TR0538A partition cannot be specified for root, special or leaf categories.

You have attempted to specify a partition level number for a category for which partitioning is not

permitted.

Ensure that you assign partition level numbers only to regular categories in levels above the lowest

levels in each dimension.

Notes

● If a partition level number is assigned to a category in a level that is excluded, cloaked, or

summarized, then that category effectively becomes a leaf category in the resulting PowerCube.

As a result, a partition level number cannot be assigned to the category.

● You can assign a partition level number to the root category of a subdimension.

Message TR0540Changing the primary drill-down from < > will cause any allocations and partitioninformation specified to be removed from the current primary drill-down. Do you wantto continue?

Transformer supports partitioning and measure allocation only in the primary drill-down path of

a dimension. If a dimension has an alternate drill-down structure, the partitioning and measure

allocation cannot be in the alternate path.

If you specify partitioning or measure allocations within the primary drill-down path, and you later

change the primary drill-down path to some other path, then Transformer must erase all of the

existing partitioning and allocation information. This will require you to re-partition your model

using levels and categories in the new primary drill-down path. Similarly, it will require you to

change measure allocation (if desired) in the new primary drill-down path.

Message TR0541A drill-down category cannot be filtered.

This message occurs when creating an MDL script that attempts to exclude a drill category. The

Transformer user interface will not allow you to exclude a drill category.

Ensure that when you reference an object in MDL, you use the correct object identifier or object

name. To view the object identifier for each object, select the Object Identifier check box in the



Titles tab of the Preferences dialog box. The object identifier will appear in the title bar at the top

of property sheet, and will also appear in a tooltip when your cursor hovers over an object. You

can also find object identifiers in the .mdl file.

Message TR0552Manual categories in this dimension can't be created because it is locked.

You can't create categories for a dimension that is locked.

To create a manual category for a locked dimension, clear the Prohibit Automatic Creation of New

Categories check box in the Dimension property sheet (General tab).

Message TR0605A column was expected but not found. Either you have referred to a column by an objectname or object identifier that Transformer is unable to locate, or you have not referencedthe column.

This error occurs when MDL syntax requires an object and is unable to find it. In this case the

object is a column. The problem may be that the column was not referenced, or that it was referenced







When a column name is changed, the object name of the column may become different in MDL

and the user interface. MDL derives the object name from the Original Name box of the Column

property sheet (General tab). The user interface takes the object name from the Column Name box,

also on the General tab.

Message TR0606The specifications for the date format and the level of detail are inconsistent. Confirmthe specifications in the column property sheet.

You have specified conflicting properties for the Date Input Format and the Degree of Detail on

the Time tab of a column's property sheet. The date format must provide sufficient information to

provide the specified Degree of Detail.

For example, if you specify a Degree of Detail of Day, (meaning that the date values represent days,

and that any measures in the input records represent daily values), then you cannot specify a date

Input Format of YM. This Date Input Format does not provide sufficient detail to populate a time

dimension with days.



Message TR0607The data class in the column property sheet is not compatible with the storage type inthe measure property sheet. Change the data class or change the storage type.

The data class of a column isn't compatible with the storage type of a measure that is based on that

column.

For example, the database stores Cost as a varchar, which is non-numeric. However, if you attempt

to change the data type in Transformer to Numeric, you will receive this error.

If your data source is an Impromptu query definition file (.iqd), you can resolve the problem in

Impromptu. In Impromptu, change the data definition for the column using the string-to-number

function. Save the Impromptu query definition file. In Transformer, remove and re-add the column.

If your data source isn't an Impromptu query definition file (.iqd), you must change the data type

for the column in the database.

Message TR0613Column <column> is referenced by name in one or more dimensions, levels, measuresor currency tables. Deleting or modifying the column name can cause these associationsto be lost. Do you want to continue?

You have attempted to delete a column that is currently being used in the model. Deleting a column

that provides category or measure values will cause the objects in the model to lose their association

with any source column.

Ensure that you don't delete columns your model requires.

Message TR0621An association was expected but not found. Either you have referred to an associationby an object name or object identifier that Transformer is unable to locate, or you havenot referenced the association.

This error occurs when Transformer's Model Definition Language (MDL) requires an object and

is unable to find it. In this case the object is an association. The problem may be that the association

was not referenced, or that it was referenced with an incorrect object name or object identifier.








Message TR0623One <association type> association from <data source name> data source is alreadydefined.

A second <association type> association from the same data source is not allowed."

This error occurs when you try to specify a second association of the same type from the same data

source to an object.

To solve this problem, either remove the association that you don't want, or make an association

to another data source.

Message TR0624One <association type> association named <association name> is already defined.

This error occurs when you try to create an association when an identical association already exists.

This error may also occur when you use an MDL script that defines an association more than once.


To solve this problem, remove the unwanted association.

Message TR0700This action will cause one cube to be deleted in the Cube Group < >. Do you want tocontinue?

You have performed some action that will cause Transformer to delete one or more cubes. This

can happen when

● you delete a category that is in a level used to define a cube group. This includes the deletion

of categories by CleanHouse.

● you create a dimension view that excludes, cloaks, or suppresses one or more of the categories

used to define a cube group.

● in the Category property sheet, you change the Inclusion setting for a category in the level that

defines the cube group to be either Suppress or Exclude.

● you exclude or suppress one of the categories that defines a cube within a cube group on the

main dimension diagram.

For example, suppose that you create a cube group based on States in a States dimension that

contains levels State and Outlet. Assume that the State level contains categories State01, State02,

and State03. Transformer will create three cubes within the cube group, one for each state. If

you then create and apply a dimension view that cloaks or excludes State01, the cube for State01

will be deleted.

You will be prompted to continue before each affected cube is deleted.



Message TR0701A PowerCube was expected but not found. Either you have referred to a PowerCube byan object name or object identifier that Transformer is unable to locate, or you havenot referenced the PowerCube.


object is a PowerCube. The problem may be that the PowerCube was not referenced, or that it was

referenced with an incorrect object name or object identifier.






Message TR0702You can't delete this PowerCube because it belongs to a PowerCube group.

Transformer will not allow you to delete a cube from within a cube group.

If you want to delete a specific cube from within a cube group, you can create a dimension view

that excludes the values associated with the cubes target category.

For more information, see "Message TR0700" (p. 324).

Message TR0705Category < > has been excluded from the PowerCube Group < >. A PowerCube can'tbe created in this PowerCube Group.

When you try to create a cube group, no cubes are created.

The problem may be that the category or level which is being used to create cube groups is sup-

pressed.

Message TR0713In Cube Group < >, one or more cubes are based on temporary key orphanages whichwill be removed. Regenerate the cube group to place the data previously in the keyorphanage cubes into the other cubes in the cube group.

During the generation of a cube group, Transformer has generated a cube that is based on a key

orphanage. Subsequently Transformer has repaired the path from the orphaned categories to an

existing source category in the target level for the cube group. As a result, you must regenerate the

cube group so that Transformer will place the orphaned data into the correct cube.

For example, assume that cube generation occurs based on three data sources: data sources A, B,

and C. A cube group is generated as follows:



● Data source A is processed, and the categories in the target level are generated as part of that

process.

● Data source B is processed, and during processing a category is encountered that does not

belong to any of the existing target categories. Moreover, no new data exists in data source B

to provide a new target category. As a result, Transformer places the orphaned category under

a key orphanage, starting at the target level, and a cube for that key orphanage is generated.

● Data source C is processed, and new data is provided that places the orphaned category under

one of the existing target categories. As a result, the key orphanage is no longer required.

Since the data from Data source B was originally placed into the orphaned cube, it is necessary to

regenerate the cube group. This will allow Transformer to place the data from data source B in the

correct cube within the group.

Message TR0716A cube that is not designated as a local cube cannot be compressed.

If you received this error message when attempting to set the location of cubes within a cube group,

the problem may be that the parent of the cube group is set to be compressed, and the default loc-

ation of the cube group is not local.

If each cube is set to a different, non-local location, Transformer may interpret the cubes to be

server cubes. The output tab of the cube within the group is pointing to the LAN location and the

Database Type setting is <From Parent>. Change the database type setting to Local and retry.

Message TR0743The currency table cannot be populated at this time because date categories have notbeen generated in the time dimension.

The currency table doesn't contain data from an external data source although you attempted to

load data.

You must generate categories before loading the external data source for your currency table.

Message TR0745A total of < > rows in the currency data were rejected because they didn't match anydate in the model.

The most common cause of this error is that the date loaded from an external data source is in a

different format from the date format set in Transformer. For example, the date format in the

external data source is YMD but the Transformer setting is DMY. It is also possible that the dates

in your currency data don't correspond with the other dates in your model.

You may need to change the date format. In the user interface, do this in the Time tab of the Column

property sheet. In MDL, this is set with the Format option in the column definition statement.



Message TR0749The currency table populate failed.

Transformer could not find the source data file. Check that the file name and path are correct. If

the source is an IQD file, check the database connection string.

Message TR0750Currency table Date column not found.

The Date column you specified doesn't exist, or you did not specify a date column.

The currency table requires a date, a rate, and one or both of a label and country code.

Message TR0751Currency table Rate column not found.

The Rate column you specified doesn't exist, or you did not specify a rate column.


Message TR0752The currency table Label or Country Code column is mandatory.

The label or country code column you specified doesn't exist, or you did not specify at least one of

them.


The Rate column you specified doesn't exist, or you did not specify a rate column.

Message TR0753The conversion rate entered is invalid. It either contains invalid characters or is lessthan or equal to zero.

Conversion rates must be positive numbers. They cannot be negative, zero, or non-numeric characters.

Message TR0800Level < > has no date function. Transformer won't generate any date categories if theyfollow a non-date level.

A non-date level exists in a time dimension. The non-date level is positioned in the dimension in

such a way that Transformer is unable to generate categories for any lower levels that may have

date functions specified.



When Transformer generates date categories, it traverses down the dimension generating date cat-

egories in levels as indicated by the date function for each level. If Transformer encounters a non

date level, the generation will stop, and Transformer will be unable to generate date categories in

the levels below the non date level.

This message occurs when you drag a column from the Data Sources list and drop it onto a time

dimension to create a new level. To fix the problem, specify a date function for the level on the

Time tab of the Level property sheet.

Message TR0802Levels in this time dimension aren't in their natural order (year, quarter, month, day).Please re-order the levels.

Some operation has made the order of date levels in a time dimension incorrect. Possible actions

that could cause this are:

● changing date functions for one or more of the date levels so that they're improperly ordered.

● reordering the date levels in the time dimension.

The date levels in a time dimension must exist in a year, quarter, month, week, and day ordering.

Re-order the levels so that they follow this pattern.

If you are working in an MDL file, you may have incorrectly referenced the level. Ensure that when

you reference an object in MDL, you use the correct object identifier or object name. To view the

object identifier for each object, select the Object Identifier check box in the Titles tab of the Pref-

erences dialog box. The object identifier will appear in the title bar at the top of property sheet,

and will also appear in a tooltip when your cursor hovers over an object. You can also find object

identifiers in the .mdl file.

Message TR0803More than 1000 date categories have been generated from your specifications.

Do you want to continue generating date categories?

You have asked Transformer to generate date categories in a time dimension. Once one thousand

categories have been generated, Transformer issues this warning so that the user can stop the gen-

eration process, possibly stopping a runaway category generation.

If you know that the time dimension does contain a vary large number of categories (such as several

years, with quarters, months, weeks, and days), then you can ignore this message. If not, then you

should discontinue category generation and verify that the time dimension is defined properly.

Check that Time tab settings on the Dimension, Drill Category, and Level property sheets.



Message TR0804The earliest date is invalid.

You have specified an invalid Earliest Date in the Dimension property sheet (Time tab).

The range of dates must be between 19000101 and 99991231 inclusive. The input format for values

in the Earliest and Latest Date boxes must be in the format YYYYMMDD.

Verify that the dates you have specified are valid, and that they fall into an acceptable range.

Message TR0805The latest date is invalid.

You have specified an invalid Latest Date in the Dimension property sheet (Time tab).

The range of dates must be between 19000101 and 99991231 inclusive. The input format for values

in the Earliest and Latest Date boxes must be in the format YYYYMMDD.

Verify that the dates you have specified are valid, and that they fall into an acceptable range.

Message TR0806The date is invalid.

You have specified an invalid date or specified a date prior to 1990 when using the Clean House

command.

The date must be later than 1989. Type a date using the format configured for your system or, in

PowerPlay version 6.6 and above, select a date by using the embedded calendar.

Message TR0807The earliest date is larger than the latest date. Make the earliest date smaller, or makethe latest date larger.

You have entered an invalid combination of Early and Late Dates in the Dimension property sheet

(Time tab).

Ensure that the Latest Date value is larger that the Earliest Date value.

Message TR0808Invalid date format.

You have attempted to specify an invalid date display format in the Dimension property sheet (Time

tab).

Ensure that the date format you have specified is a valid one.



Message TR0809The current period can't be a special category.

You have attempted to set the current time period to the value represented by a special category.

You cannot use a special category to set the current period. Use a regular category to set the current

time period.

Message TR0810The date levels specified for this dimension aren't allowed. Ensure that lunar and calendardate level functions haven't been combined. Refer to the online Help for more informa-tion.

You have combined both calendar and lunar levels within the same drill-down path of a dimension

that contains an alternate drill-down structure.

When specifying date level functions in a time dimension, lunar time periods and calendar time

periods cannot be combined within a single alternate drill-down. For example, it is invalid to have

lunar years with calendar months in the same drill-down path. Within a single alternate drill-down

path, all date functions must be exclusively lunar periods or calendar periods.

In addition, the only time period at which alternate drill-downs of type Calendar and Lunar can

converge is the Day level. You cannot set up a Calendar drill-down that joins with a Lunar drill-

down at the Year, Quarter, Month, or Week level.

Restructure your time dimension so that date level functions are specified correctly.

Message TR0811The date specified by the year's start-day doesn't match the weekday specified by theweek's start-day. Both dates must fall on the same day of the week.

You have created a lunar drill-down in which the date on which the year begins does not match

the day on which the week begins. Because lunar years contain 52 weeks, the date on which the

year begins must match the week day designated by the week start day.

For example, if you create a drill-down path that uses lunar time periods, and you specify that the

Year Begins property is 19960101, then the Week Begins On property must be set to Monday, as

January 01, 1996 is a Monday.

Verify that the Week Begins On and Week Begins On properties coincide in the Drill Category

property sheet (Time tab).



Message TR0812<level_name> is not a valid convergence level in this time dimension. Please refer tothe online Help for information on alternate drill-down paths in time dimensions.

The current convergence level does not share the same set of categories between the date drill-down

paths for one of the following reasons:

● lunar and solar drill-downs are combined

● existing drill specifications make this convergence level illegal

Message TR0813The current period for dimension <dimension_name>wasn't changed because category<category_code> wasn't found.

You have attempted to change the current date in batch mode using the -t command line option.

Transformer was unable to find the category code specified as part of the -t command line option.

Verify that you specified the correct category code for the category you want to use to set the current

date.

Message TR0815The interval between the year start-days for alternate drill-downs <drill-down_name1>and <drill-down_name2> in dimension <dimension_name> must be equal to one ormore whole units of time as represented by the convergence level. Refer to the onlineHelp for more information.

You have created an alternate drill-down structure in the time dimension and specified incompatible

dates for the year start days in each drill-down path. Ensure that the start dates for the years in

each drill-down path are the same.

Message TR0816A signon was expected but not found. Either you have referred to a signon by an objectname or object identifier that Transformer is unable to locate, or you have not referencedthe signon.


object is a signon. The problem may be that the signon was not referenced, or that it was referenced









Message TR0817The signon may not be deleted or have its name changed because it is associated witha data source.

You cannot change the name of the Signon in Transformer. The name is taken from the COGCON-

NECT.ini database definition and Transformer uses that name when connecting to the database.

Message TR0900There were changes to the (default) earliest date and latest date. This will cause datecategories to be deleted from this dimension.

You have made changes to the range of dates Transformer will use when generating date categories.

Ensure that the change was intended. If you accept the change, all of the categories in your time

dimension will be deleted.

Message TR0904The date level settings could generate in excess of < > categories. Do you want tocontinue?

Transformer is warning you that it may generate an excessively large number of date categories.

This can happen when you have specified a relatively wide range of dates to include in the generation

of date categories, and the time dimension includes low levels, such as Day.

In cases where your model requires all of these dates, you can ignore this message. However, if you

really only need two or three years of data, narrow the range in the Time tab of the Dimension

property sheet.

Message TR0906The associated column that is specified for dimension < > can't be found in the DataSources list.

Transformer is unable to find a column to match the one specified as the column associated with

a regular time dimension. This can happen when

● the source data file for a data source changes

● you delete or change the name of a source column

● you change the column name specified in the Dimension property sheet (Time tab) for the time

dimension

Ensure that the Column name specified in the Time tab of the Dimension property sheet matches

a column in one or more of your data sources.



Message TR0907A dimension was expected but not found. Either you have referred to a dimension byan object name or object identifier that Transformer is unable to locate, or you havenot referenced the dimension.


object is a dimension. The problem may be that the dimension was not referenced, or that it was







Message TR0914This dimension contains an alternate drill-down path with a convergence level. All drill-down paths that share this convergence level must use the same calendar. A mixture ofcalendar and lunar date functions isn't possible.

In a time dimension, you have attempted to set the convergence level for alternate drill downs of

type Calendar and Lunar to some level other than Days. You can only join alternate drill downs

of different types at the Days level.

Message TR0917Please specify a column for a regular time dimension.

You have attempted to create a Regular time dimension without providing the name of a column

that provides date values for the categories in the dimension.

In the General tab of the Dimension property sheet, specify the name of a data source column that

provides date values for the dimension.

Message TR0919Measure <measure_name> is an allocated measure in dimension <dimension_name>.Cube group <cube_group_name> is based upon this dimension. This measure will notappear in the .mdc unless an external detail level is specified for the cube group.

When an allocated measure is used in a PowerCube that is built as part of a cube group, the cube

group must be defined with an external level of detail. The external level of detail ensures that

measure values required to perform the allocation are available in the PowerCube.



Message TR1003The Impromptu Query Definition contains a syntax error.

You have attempted to use a data source of type Impromptu Query Definition (IQD) and the contents

of the IQD file have become corrupted.

Verify that you are using the correct IQD file. If possible, recreate the IQD file using Impromptu.

Message TR1005IQD Data Source file <file_name> contains one or more BLOB data columns. BLOBdata columns cannot be handled by Transformer.

Transformer does not support BLOB data types.

Recreate the IQD file using Impromptu, and omit the column or columns with the BLOB data types.

Message TR1100The import file < > has an invalid keyword on line < >.

There is an invalid keyword in the MDL script. Check that all the keywords are spelled correctly

on the specified line.

Message TR1101Transformer detected a syntax error at line < > in file < >.

There is an error in the MDL script. For details of the correct syntax of MDL statements, see the

Transformer MDL Reference, which is available in English.

Confirm that the data contained on the line referenced in the error message does not contain erro-

neous or invalid data, or corruption.

Try a test model which has this line remarked out. In MDL, delete the line.

This error message is generally followed by one or more additional error messages. For help on

those messages, refer to the online Help. In the Help menu, choose Contents and Index, and then

choose the Find tab. Search for the message by its TR number; for example, TR1101.

Message TR1102Transformer detected a model error at line < > in file < >.

There are several conditions that could cause this error. Usually, the conditions are explained in

more detail in additional error messages that follow in the same dialog box.

This error message is generally followed by one or more additional error messages. For help on

those messages, refer to the online Help. In the Help menu, choose Contents and Index, and then



choose the Find tab. Search for the message by its TR number; for example, TR1102. For information

about MDL, see the Transformer MDL Reference, which is available in English.

Message TR1106The object for the object ID given on line < > is the wrong type in file < >.

This error message is generated when an object identifier specified in the MDL file exists, but is

being applied to the wrong object type.

For example, 117 is the object identifier for a level. In the MDL file, a category is identified with

the object identifier 117.

To solve the problem, use a different object identifier. Object identifiers can be up any number

greater than 100 and less than 4,294,967,296.

Message TR1109The import file < > contains a string on line < > which contains a new line character.

Check that there is no newline character on the indicated line in your data source file.

If this does not work, save the model as .mdl, open it and retry.

Message TR1304The associated column you specified for measure <measure_name> can't be found inthe Data Sources list.

In defining the measures for your model, you have inadvertently specified the name of a column

that does not exist.

Ensure that the name of the column you specify for a source measure matches the name of a column

in the Data Sources list.

Message TR1307This measure is used as a weight by measure <measure_name>.

You have attempted to delete a measure that is used as a weight for a Regular, Time State, or

Duplicates rollup.

Before you can delete the measure, you must ensure that it is not being used to weight an average.



Message TR1308This measure is referenced by calculated measure '<measurename>'. Please redefinethe calculated measure.

This error message is generated when an attempt is made to delete a measure that is referenced by

another measure. Confirm that the measure is not part of a calculation or required by another

measure in the model.

Message TR1309The calculated measure can't use Before Rollup because it refers to a measure with theAfter Rollup setting.

You have attempted to set the rollup timing for a calculated measure that references a measure with

a conflicting setting for the timing of rollups.

For example, if you create the calculated measure M1 with a Regular Timing of After Rollup, you

cannot then reference measure M1 in the expression for the calculated measure M2 if the timing

for M2 is set to Before Rollup.

Message TR1310A measure was expected but not found. Either you have referred to a measure by anobject name or object identifier that Transformer is unable to locate, or you have notreferenced the measure.


object is a measure. The problem may be that the measure was not referenced, or that it was refer-

enced with an incorrect object name or object identifier.






Message TR1312The type of Time State Rollup used must be the default because there are multiple timedimensions.

Your model contains more than one time dimension, and when setting the type of time state rollup,

Transformer has detected rollups other than the default.

Transformer cannot perform time-state rollups other than default when there is more than one time

dimension in the model. If a time state rollup other than default is required, you must remove one

of the time dimensions from the model.



Message TR1319Measure <measure_name> can be in multiple data sources only if the lowest levels ineach regular dimension that are associated with the data sources are identical.

This error message is generated when Transformer detects a measure that obtains source values

from columns in two or more data sources, but the data sources are not associated with the same

levels and dimensions.

For example, you have two data sources that contain source values for the Revenue measure. The

data sources are not related in any other way. Both data sources have columns that are used as

source values for levels in different dimensions. Transformer can't perform rollups correctly because

it doesn't know how to associate measure values with specific categories.

In practical terms, this means that you should only obtain measure values from multiple data sources

if all of those data sources contribute to the same dimension.

To solve the problem, there must be a relationship between multiple data sources for the measure

to apply properly to your dimensions. Reconstruct your data sources so that the measure values

are obtained from a single data source, or so that the data sources that provide measure values

contribute to the same dimension. Use the Show Scope command (Edit menu) to check if data

sources contribute to a dimension.

Message TR1320Data sources associated with measure <measure_name> include levels only in thealternate drill-downs and not in the primary drill-down for dimension<dimension_name>.No source values will be in the cube for this measure.

In constructing the data sources for your model, you have created a transactional data source

(containing measures) that provides category values only within an alternate drill-down path.

In order for measure values to be written to the PowerCube, they must be associated with a

dimension's primary drill-down path. You can check the scope of the measure by using the Show

Scope command.

To resolve this problem, you must reconstruct the measure's source data file so that it contains

columns which provide category values for levels in the primary drill-down.

Message TR1340The measure ' ' can use 'Ignore missing value in Time state rollup' only if it has Timestate rollup of Average, Weighted Average, Minimum or Maximum.

This error typically occurs when an unsupported rollup type such as First Period, Last Period, or

Current Period, is specified prior to setting the IgnoreMissingValue flag to TRUE. The default, and

the only measure-handling supported in versions of Transformer earlier than Series 7 Version 4, is

FALSE: retain null and missing values for all time state rollup types.

For more information about rollup types, consult the index for the online help or the MDLReference

documentation.



Message TR1502A data source was expected but not found. Either you have referred to a data source byan object name or object identifier that Transformer is unable to locate, or you havenot referenced the data source.


object is a data source. The problem may be that the data source was not referenced, or that it was







Message TR1503Choosing the 'Maximize Data Source Access Speed' option may increase processing speedin some cases but Transformer will no longer detect multiple instances of source valuesin levels marked as Unique. Neither will it do unique moves.

If you choose the Maximize Data Access Speed option for a data source, processing speed may

increase. However, Transformer won't detect any uniqueness problems in this data source. Neither

will the unique move feature work for this data source.

If you aren't sure that your data is unique in your data source when you have declared it to be

unique, or if you use the unique move feature for this data source, select the Verify Category

Uniqueness option.

Message TR1601Transformer couldn't open the model file.

The .py? or .qy? file that you are attempting to open is not valid.

Message TR1700The category <category_name> in the dimension <dimension_name> has < > imme-diate descendants. This number should normally be close to 7. More levels may beappropriate in this dimension.

This is a design recommendation and will not affect your ability to generate categories or create a

PowerCube.

Categories with too many immediate descendants will result in PowerPlay displays which are difficult

to analyse. For example, a time dimension which has only two levels, Years and Days, would result

in Year categories with 365 immediate descendants. This could be very difficult to analyse in



PowerPlay. Adding more levels, such as Quarters and Months, improves the organization and

usability.

You can decrease the number of immediate descendants by adding manual levels to your model.

You can control when you receive this message by changing the number in the Warn When the

Number of Child Categories Exceeds box, Preferences dialog box (General tab).

Message TR1703A data input conversion or overflow error occurred at source record number < > formeasure < > in source file < >.

This error message may be caused by an incorrect data type for the measure. Confirm that the data

type is valid for the values stored within it and retry.

Message TR1900The wrong ID stamp was detected on<cube_name>. This often happens when the modelisn't saved after creating a PowerCube.

Transformer has detected an inconsistency between the model and the PowerCube that you are

attempting to update.

This can occur when you attempt to update an incrementally updated PowerCube using a version

of that cube other than the one the model was most recently saved with. For example, if you save

a model after creating a PowerCube using February's data, you cannot restore the January version

of the PowerCube without raising this error.

By clicking Cancel when this error occurs, you set the PowerCube status to Invalid. You can the

Adopt the cube, in which case Transformer resets the status to OK. If you click OK when this error

occurs, Transformer will proceed with the incremental update.

Message TR1903Incremental update of <cube_name> cannot proceed because of the <status> statusof an existing PowerCube.

Transformer has detected a condition that prevents it from proceeding with an incremental update

to a PowerCube.

This is caused when an incremental update of a PowerCube fails for some reason. For example, if

you attempt to incrementally update a PowerCube when that cube is being viewed in PowerPlay,

then the update will fail.

When Transformer fails to incrementally update a PowerCube, it assigns that PowerCube a status

of FAILED. If you then attempt to recreate the cube, Transformer will not allow you to do so,

because it is possible that the cube already contains some of the data from the most recent increment.

Reprocessing that increment could result in some data being added to the cube twice.

To recover from the problem, you can



● revert to a backed up copy of the model (prior to processing the most recent increment) and

associated PowerCube, and then attempt to perform the incremental update again.

● revert to a backed up copy of the model (prior to processing the most recent increment) without

a backed up copy of the PowerCube, and then attempt to perform the incremental update again.

When you do this, Transformer will issue a warning message indicating that the time stamps

on the PowerCube and the Model don't match. If you proceed to update the PowerCube, you

run the risk of duplicating some data in it.

● rebuild your cube using the combined data from all increments. Once the cube has been recre-

ated, you can proceed with new increments in the normal fashion.

If you know that the failed cube was caused because the cube was opened by another application,

you can adopt the existing cube and then rerun the incremental update.

Message TR1907Transformer can't gain access to database < > with signon information < >.

If you are running server Transformer, ensure that the Prompt for Password box is cleared (Signon

property sheet, General tab). Prompting for a password is not relevant for server Transformer.

The cube may be successfully created, even though this error message is generated.

Message TR2000The main dimension is required in the PowerCube Group.

In a PowerCube Group, you have attempted to omit the dimension containing the level that defines

the group.

You cannot omit this dimension. If you want to omit a portion of the dimension, use dimension

views. Be aware, however, that omitting one of the categories that defines a cube within the group

by excluding or cloaking will cause that cube to be deleted from the group.

Message TR2001The main dimension is required in the Time-Based Partitioned Cube.

In a time-based partitioned cube, you have attempted to omit the dimension containing the level

that defines the cube.


views.



Message TR2002The main dimension is required for the User Class.

In a User Class that has been added to a time-based partitioned cube, you have attempted to omit

the dimension containing the level that defines the time-based partitioned cube.


views.

Message TR2306You have specified a label column. Please specify an associated column also.

You may have deleted the Associations from a Level property sheet. Confirm that this information

is complete.

Message TR2307A level was expected but not found. Either you have referred to a level by an objectname or object identifier that Transformer is unable to locate, or you have not referencedthe level.


object is a level. The problem may be that the level was not referenced, or that it was referenced







Message TR2308A date function has been chosen but the dimension is not a time dimension.

Date functions are set for the dimension, which is not a time dimension.

The dimension type, which is set on the General tab of the Dimension property sheet, is not Time.

Or in MDL, the option DimType Date is not included in the Dimension definition statement.

At the same time, a date function has been set for the dimension. Date functions are set on the Time

tab of the Dimension property sheet. Or in MDL, they are set with the options DaysInWeek,

EarliestDate, LatestDate or ManualPeriods.

Either set the dimension as a time dimension or delete the date functions.



Note: If you are deleting date functions in the user interface, you may need to temporarily select

Dimension Type Time on the General tab of the Dimension property sheet so that the Time tab is

enabled.

Message TR2312You made level < > into a convergence level that connects two or more alternate drill-down paths. Categories in convergence levels must have unique source values. Can therebe two categories in this level with the same source value?

You have created a convergence level but have not designated it as unique. Uniqueness is set by

selecting the Unique check box in the Source tab of the Level property sheet. Or in MDL, it is set

by the option UniqueCategories True in the level definition statement.

If you answer No to this question, the setting will be changed to Unique and the convergence level

will be created.

If you answer Yes to this question, the convergence level will not be created.

Message TR2313You are creating a level as a convergence level that connects two or more alternate drill-down paths. Categories in convergence levels must have unique source values. Is thislevel unique by this definition?

You have created a convergence level but have not designated it as unique. Uniqueness is set by

selecting the Unique check box in the Source tab of the Level property sheet. Or in MDL, it is set

by the option UniqueCategories True in the level definition statement.

If you answer Yes to this question, the setting will be changed to Unique and the convergence level

will be created.

If you answer No to this question, the convergence level will not be created.

Message TR2314This is not an unique level.

This error message indicates that you have a problem related to level uniqueness. This can occur

when

● you disable some of Transformer's error-checking options so that it no longer verifies that

unique levels contain unique categories

● you use multiple data sources in a model and don't define levels associated with columns from

multiple data sources as unique


Transformer asks you to verify the uniqueness property before proceeding

● you specify that a level is unique and Transformer detects that it is not



The error message may also be due to a data problem. Confirm that your source data file is valid

for the model.

Message TR2316The level <level_name> can't be deleted because it is referenced by PowerCube<cube_name>.

You have attempted to delete a level that defines a PowerCube group. Transformer will not allow

you to delete such a level.

If you want to delete the level, you must first delete or modify the PowerCube group that uses the

level's categories to define cubes within a cube group. For more information, see "Define PowerCube

Groups " (p. 130).

Message TR2317The level <level_name> is designated as unique.

Source value <source_value> was used in an attempt to create a category in the path<drill_path_1>.

< source_value > already exists in level <level_name> in the path <drill_path_2 >.


when









for the model.

Message TR2318Transformer has detected < > attempts to create a category in more than one path.

Refer to the online help for a detailed explanation of level uniqueness.


when











for the model.

Message TR2319A convergence level in an alternate drill-down path must be unique.


when









for the model.

Message TR2320You are attempting to sort a date level in descending order. This will create incorrectrelative-time calculations. Do you want to continue?

Transformer uses the ordered set of categories in a time dimension to perform relative time calcula-

tions for relative time categories. If you reverse the order of the categories in the time dimension

by sorting them in descending order, Transformer will no longer be able to perform relative time

calculations correctly.

When re-ordering categories in the time dimension, you must decide which aspect of the model is

most important: the order of categories within the time dimension, or the relative time categories

that are included within the time dimension.



Message TR2321A share object level must be an ancestor of a share target level.

You have tried to apply a share object to a category or categories at one level when the share object

that is currently set is

● at a lower level than the categories to which you are applying the share object, or

● not an ancestor of the categories to which you are applying the share object.

Ensure that when you apply a share object, you have set the share object to be an ancestor category.

Message TR2322A share object level must be in the same dimension as its target categories.

You have tried to apply a share object to one or more categories when the share object is currently

set to a category in another dimension. This is not supported.

A share object must be an ancestor category of the category to which you are applying the share

object.

Message TR2323The share object ID is invalid. A share object must be an object identifier of a categoryor level in the same dimension.

You have attempted to specify an invalid share object. The share object

● must be an existing category or level

● must be at a higher level in the same dimension

● must be specified by its object identifier

To view the object identifier for each object, select the Object Identifier check box in the Titles tab

of the Preferences dialog box. The object identifier will appear in the title bar at the top of property

sheet, and will also appear in a tooltip when your cursor hovers over an object. You can also find

object identifiers in the .mdl file.

Message TR2324Partitioning can be specified only for levels in the primary alternate drill-down.

Typically, this error message is written to the log file in response to an MDL script that attempts

to specify a partition number for a level in an alternate drill-down path. The user interface for

Transformer does not normally allow you to specify partition level numbers in levels other than

those in the primary drill-down structure.



To rectify the problem, you should verify that your partitioning is specified for levels in the primary

drill-down only. If partitioning is required for levels in an alternate drill-down path, consider

making that path the primary drill-down path. Note, however, that changing the primary drill-

down path will remove all current partitioning information from your model.

Message TR2325The Short Name column you specified for level <level_name> isn't in the Data Sourceslist.

You have specified a column to supply Short Names for the categories in a level, and Transformer

is unable to find the column that you specified.

Ensure that the column name you specify matches one that is defined in a data source for your

model.

Message TR2326The Description column you specified for level <level_name> isn't in the Data Sourceslist.

You have specified a column to supply a Description for the categories in a level, and Transformer

is unable to find the column that you specified.

Ensure that the column name you specify matches one that is defined in a data source for your

model.

Message TR2502A view was expected but not found. Either you have referred to a view by an objectname or object identifier that Transformer is unable to locate, or you have not referencedthe view.


object is a view. The problem may be that the view was not referenced, or that it was referenced









Message TR2503Special categories can't be summarized in dimension views.

This error message typically appears in the log file in response to an action requested by an MDL

script. Under normal circumstances, the Transformer user interface will not allow you to perform

invalid operations on special categories in a dimension view. You cannot exclude, summarize, or

cloak a special category in a dimension view.

If you encounter this error, ensure that the category object identifier that you specified in the

CatUpdate or CatMake statement does not refer to a special category.

Message TR2504Special categories can't be suppressed in dimension views.







Message TR2505Special categories can't be cloaked in dimension views.







Message TR2506An apex action can't be done with special categories.









Message TR2507This dimension view can't be deleted. It is pre-defined.

You have attempted to delete one of the following predefined dimension views:

● All Categories

● Omit Dimension

These default dimension views are created for every dimension, and are used to include or exclude

entire dimensions from a PowerCube. You cannot delete, reorder, or rename them.

Message TR2508This view is in use. If it is removed, references to it will use All Categories. Do you wantto remove it?

You have attempted to delete a dimension view that is currently being applied to one or more of

the cubes in the PowerCubes list. By deleting the view, you remove any restrictions (excludes,

summaries, cloaking, and so on) that were imposed on those cubes by the view. This can be an

issue if you are creating cubes for specific audiences, and the removal of the view will allow some

users to view data they should not be able to access.

MessageTR2509The position of the pre-defined views 'All Categories' and 'Omit Dimension' can't bechanged.

You have attempted to reorder one of the following predefined dimension views:

● All Categories

● Omit Dimension



Message TR2510The names of the predefined views 'All Categories' and 'Omit Dimension' can't be changed.

You have attempted to modify one of the following predefined dimension views:

● All Categories

● Omit Dimension





Message TR2600An error occurred during the creation of a PowerCube. Try creating the PowerCubeagain.

There is a problem with the temporary files that are created during the creation of a PowerCube.

The location of the temporary files is set on the Directories tab of the Preferences dialog box, Data

Temporary Files (dir1;dir2). If the location box is blank, use the What's This? help on the property

sheet for information about default locations.

You may have insufficient disk space to create these files. Ensure that there is sufficient free space

in the file location.

If this is not the problem, try deleting all temporary files that are currently in existence. They may

have become corrupted.

Message TR2601An error occurred while saving.

Transformer was unable to save the model.

Verify that the directory in which you are saving your model is one to which you have write access.

If a model file with the same name already exists, verify also that the existing .PY? file you are

attempting to overwrite is not read-only.

Message TR2605Transformer can't process this model. It uses features that are part of a more compre-hensive edition than the one you have installed. Refer to the documentation for adescription of the 4 available editions.

This error message is returned when a model using advanced features of PowerPlay Transformer

is opened in an edition that does not support them.

PowerPlay is available in several distinct editions. The differences between these editions are in the

scope of what you can do in Transformer.

Additionally, a server model can only be opened in server Transformer.

Message TR2606Transformer has detected one or more suspended models.

Show the list of suspended models?

Transformer has detected suspended models (.qy? files) that contain checkpoints of suspended

models. You can examine the list of suspended models and then attempt to recover them.

The recovery dialog box offers two options:



● Open Suspended Model at Last Checkpoint opens the selected model at the last checkpoint

before failure. You can then view the failed model, continuing to develop it from that checkpoint.

● Open Last Saved Model and Delete Checkpoint File opens the selected model as it was last

saved, and then deletes the corresponding .qy? file. If the model was not previously saved,

Transformer will not be able to open it, but you can then choose to retain the .qy? file to open

the suspended model at the last checkpoint.

Message TR2608Suspended models were not found.

You have asked Transformer to view any suspended models that may exist, and Transformer cannot

locate any suspended models.

By default, Transformer stores suspended models with the extension .QY? in your temporary dir-

ectory . If you know that suspended models should exist (because Transformer failed at some point),

search your path for files with this extension.

Message TR2700The columns in formulated data source <data_source_name> don't match the originalsource columns.

One of the source data files associated with a data source has changed so that its columns no longer

match the ones in the data source object defined for the model.

When you create a model, the columns in each data source are saved as part of the model definition.

If you reorder, add, delete or rename the columns in the source data file, Transformer detects that

the columns in the source file no longer match those in the data source.

To solve this problem, use the Modify Columns command to match the columns in the source data

file with the columns in the data source.

Steps

1. Click the affected data source.

2. From the Tools menu, click Modify Columns. A plus sign (+) appears to the left of all columns

in the model for which Transformer detects a change.

3. For each changed column, click Match, Add or Remove.

Message TR2701The columns specified by the alternate data source for PowerCube <cube_name> don'tmatch the columns in <data_source_name>.

The columns in an alternative source data file specified for a PowerCube in its property sheet

(General tab) do not match the columns in the data source defined in the model.



Ensure that when you use an alternative data source to create a PowerCube, the columns in the

alternative file match the columns in the data source that the alternative file is associated with.

Message TR2702There isn't a data source that is associated with dimension < >.

The specified dimension has no data sources associated with it.

To resolve this problem, ensure that

● you have included all of the data sources required for each dimension in your model

● the column names for each level match the name of a column in the Data Sources list

● you have not created a dimension that contains only manual levels

● you have not deleted a data source required for a dimension

● you have not set the Timing property of the data source such that the data source is never used

to generate categories for the model.

Message TR2703A column in data source <data_source_name > matches level <level_name> indimension <dimension_name> by name. Values from this data source column cannotbe related to categories in this level because insufficient context is specified in the datasource.

Transformer issues this error message when generating categories for the model if it is unable to

associate a data source column with some level in the model.

● You can solve the problem in the following ways:

● If the level you want to associate with the column contains only unique category values, select

the Unique check box on the Level property sheet (Source tab). This allows Transformer to

directly associate the values from the column with the level. Transformer can then use columns

from other data sources to obtain values for the ancestor categories.

● Restructure your source files so that the data source named in the error message contains all

of the required ancestor columns.

For all ancestor columns, ensure that the Data Class box in the Column property sheet (General

tab) is not set to Ignore.



Message TR2704Dimension <dimension_name> contains an unbalanced category-tree in which not allleaf categories exist in the same level.

Transformer has encountered a condition in which not all leaves exist at the same level. This can

occur when your model uses multiple data sources and those data sources generate categories for

disjoint levels.

For example, suppose two data sources are used to generate categories for a Products dimension

that contains the levels Product Line, Product Brand, and Item. If one data source contains values

that generate categories down to the Item level while the other contains values that generate categories

down only to the Product Brand level, then the dimension is unbalanced, as leaves occur at different

levels in the dimension hierarchy.

Ensure that all data sources generate categories to the same level in the dimension. Also, ensure

that the timing for each data source in the Data Sources list is set so that categories are generated

using the appropriate data sources.

Message TR2705PowerCube <cube_name> doesn't have any measures included.

You have excluded all of the model's measures.

Open the property sheet for the PowerCube you are creating, and click the Measures tab. Ensure

that you have included at least one measure.

Message TR2706Level <level_name> can't be reached in drill-down path <drill_category_name> fordata source <data_source_name>. A level can be the first source level, be directlybelow a source level, or be designated as a unique level. All measures in data source<data_source_name> will have zero values in the PowerCube.

The level that is identified in the error message is not accessible in the listed drill-down path.

If the level is associated with a column in a data source that does not contain the ancestor levels,

ensure that the level is unique. Ensure also that identically-named column(s) exist in one or more

other data sources, and that these other data sources also include the columns that generate categories

for ancestor levels.

Message TR2707Measure<measure_name1> is referenced by the calculatedmeasure<measure_name2>.These measures must have the same rollup timing.

You have specified calculated measures, one of which references the other, and which have conflicting

rollup timings.



For example, you have created calculated measure M1 with a Regular timing of Before Rollup. If

you then create calculated measure M2 with a Regular Timing of After Rollup, and you reference

M1 when defining M2, the rollup timings will conflict.

Ensure that the rollup timings specified for calculated measures do not conflict.

Message TR2708Regular measures <measure_name1> and <measure_name2> are referenced by thecalculated measure <measure_name3> , which has a rollup timing of 'before rollup'.Therefore, both regular measures must be associated with the same data sources.

When you reference two or more measures in a calculated measure with a rollup timing of Before

Rollup, ensure that the measures come from the same data sources.

This is required because when rollups are performed, each component parts of the calculated

measure must be available to provide the necessary values to perform the rollup. This is only possible

when the component parts of the calculated measure exist in a single data source. To resolve the

problem, restructure your source data file so that all source measures used in a "Before Rollup"

calculated measure are contained in the same file.

Message TR2710Level <level_name> has an allocation specified for measure <measure_name1> usingmeasure <measure_name2>. The referenced measure must include a deeper level inits measure scope than the measure being allocated.

A measure that you have used to allocate another measure does not cover the portion of the

dimension map over which you are attempting to allocate values.

For example, you cannot proportionally allocate values for Measure A to lower levels in a dimension

using Measure B if Measure B's domain does not include the levels to which you are attempting to

allocate.

Message TR2711There are no data sources set for category generation.

You have attempted either to generate categories or create PowerCubes, and no data sources are

set to be run as part of the category generation process.

To generate categories, open the property sheet for each of the data sources in your model. On the

General tab, ensure that the Generate Categories check box is selected for each data source that

provides category values for the model.



Message TR2712There are no data sources set for PowerCube creation.

You have attempted to create a PowerCube using either the Create PowerCubes command or

PowerBar button. However, no data sources are set to be run as part of the PowerCube creation

process.

To create PowerCubes, open the property sheet for each of the data sources in your model. On the

General tab, ensure that the PowerCube Creation check box is selected for each data source that

provides PowerCube information (measure values) for the model.

Message TR2713A circular allocation reference has been detected for measure <measure_name1> inlevel <level_name>. Measure <measure_name1> is allocated using another measure.At the same time another measure is allocated using <measure_name1>.

Measures cannot be allocated by their own values, or by other measures that are allocated by those

values.

For example, if you allocate Measure A using values for Measure B, you cannot then allocate some

other measure (say, Measure C) using values for Measure B if Measure C is allocated using Measure

A values.

Message TR2714A circular allocation reference has been detected for measure <measure_name1> incategory <category_name>. Measure <measure_name1> is allocated using anothermeasure. At the same time another measure is allocated using <measure_name1>.

Measures cannot be allocated by their own values, or by other measures that are allocated by those

values.

For example, if you allocate Measure A using values for Measure B, you cannot then allocate some

other measure (say, Measure C) using values for Measure B if Measure C is allocated using Measure

A values.

Message TR2715Measure <measure_name1> is allocated by the calculatedmeasure <measure_name2>.Measure <measure_name1> is a component of the calculated measure.

A calculated measure references a measure that uses that calculated measure for allocation.

For example, if the measure Fixed Costs is used in the expression for a calculated measure Fixed

Cost Ratio, as in

Fixed Cost Ratio = (Fixed Costs/Sales) * 100

then you cannot use the measure Fixed Costs to allocate values for the measure Fixed Cost Ratio.



Message TR2716Category <category_name> is inaccessible in dimension <dimension_name>. No datacan be reported against it in the PowerCube. A category is accessible if it or a descendantexists in a key level, or it has an ancestor with allocated measures values.

This message is simply a warning.

You have created a dimension in which one or more categories cannot be accessed. This can occur,

for example, when you create a special category that is not connected to a source category. Another

cause is that there is no Current Period category: the Automatically Set the Current Period check

box is cleared (Dimension property sheet, Time tab, for a time dimension) and no date category

has been selected as the current period (Diagram menu, Set Current Period).

Message TR2717Category <category_name> has no children in dimension <dimension_name> in atleast one dimension view or Cube Group. This category cannot have a partition specified.

Check Model has detected a condition in which a dimension view has caused one or more categories

to become the lowest-level active categories in a dimension. In essence, these categories become leaf

level categories. Because Transformer does not allow partitioning on leaf level categories, this is

not allowed.

To remedy the problem, you can either

● repartition your model so that the categories that become leaves in the dimension view are not

partitioned. This will require you to completely repartition the model.

● change the dimension view so that the partitioned categories are not at the lowest level in the

view

Message TR2718Dimension < > has no measures associated with it. During PowerCube generation, nodata source will be able to provide data for this dimension.

In a model that reads data from multiple data sources, the data sources that are used to provide

category values for the named dimension have no measures associated with them.

A possible solution is to ensure that when you set up multiple data source models, you include data

sources that contain measures for each dimension. For multiple data source models, this requires

you to set up

● a structural data source that provides category values for the dimension

● one or more transactional data sources that contain one or more columns with measure values

● at least one column associated with a unique level in the dimension



Message TR2719Data source < > contains no measures. It is referred to as a structural data source andshould come before all measure data sources in the data sources list.

Transformer assumes that data sources without measures are intended to provide structure for the

model by generating categories. As a result, Transformer processes these data sources prior to

processing those with measures. In order to keep the data sources in the Data Sources list sorted in

the order that Transformer processes them, you should place structural data sources at the top of

the Data Sources list with transactional data sources below them.

Message TR2720Measure < > has a time state rollup specified. For this reason, no categories may besummarized or cloaked in time dimension views. Category <> is summarized or cloakedin at least one view in dimension < >.

Transformer does not allow summary dimension views in a time dimension when a time state rollup

has been specified for one or more measures.

To resolve this problem, you can do one of the following:

● eliminate the summary view from the time dimension

● eliminate the time-state rollup for the measure

Message TR2723A gap has been detected between partition<n> and<n>. No gaps in partition numbersare permitted.

You have attempted to specify partition level numbers for a model, and you have inadvertently

missed a number in the specification of partition level numbers.

For example, if a dimension contains 4 levels (Level1 through Level4), then you cannot assign

partition level 1 to higher-level categories in the dimension and then assign partition level 3 to

descendants. Partition level 2 must be assigned to intermediate descendants (between the categories

assigned levels 1 and 3).

The same holds when you specify partition level numbers across dimensions. You cannot set level

1 for categories in a dimension and level 3 for categories in another dimension without setting level

2 somewhere else.

Message TR2724Partition <n> is defined in dimensions <dimension_name1> and<dimension_name2>.A partition number cannot be assigned in more than one dimension.

When you assign a partition level number, that number can only be assigned to categories within

a single dimension.



For example, you cannot include categories from both a time dimension and a Products dimension

to partition level 1.

Check the partition numbers that are currently specified for the categories in your model.

Message TR2725Category <category_name1> is a descendant of category <category_name2> and sharethe same partition number in dimension <dimension_name>. A partition number canonly occur once in a category path of a dimension.

When you specify partition level numbers, ensure that you don't use the same level number more

than once in the path from the root to the leaf of a dimension.

For example, in a time dimension, it is valid to assign partition level 1 to the years 1990 through

1995, and then use the same level number (1) for quarters in 1996. However, once you have assigned

level 1 to a year, you cannot then assign the same level number to any category in any lower level

of that year.

Message TR2726Category <category_name1> is a descendant of category <category_name2> indimension <dimension_name>. The partition number specified for <category_name2>cannot be greater than the partition number of <category_name1>.

When you specify partition level numbers in a dimension, the numbers of lower levels must be

greater than those specified for their ancestors.

For example, in a time dimension, if you assign partition number 1 to categories in the Year level

and number 2 to categories in the Month level, you cannot then assign partition level 3 to categories

in the Quarters level.

When you set up partition levels, ensure that you start numbering at 1 for higher levels, 2 for child

categories, 3 for deeper child categories, and so on.

Message TR2727Measure <measure_name1> is an allocated measure which is referenced by the beforerollup calculated measure <measure_name2>. A before rollup calculated measurecannot have allocated measures as components.

When setting up a calculated measure with a rollup timing of Before Rollup, you cannot use an

allocated measure in the expression for the calculated measure.

Because allocation is performed at run time, the allocated value is not available to perform the

calculation before rollups are performed.



Message TR2728Data Source <data_source_name> is an .iqd definition which references database<database_name>. This database has no database type specified.

Transformer is unable to locate a database type (for example OR for Oracle or CT for Sybase Client

Library).

Check the database definition specified in Impromptu, as well as the entry for the database in the

[PowerPlay Server List] section of your Cognos.ini or COGCONNECT.ini file.

Message TR2729Measure <measure_name> has a time state rollup specified. This measure does nottouch the lowest level in dimension <dimension_name> causing time state rollup valuesto be zero in any cubes that reference this measure.

When defining time-state measures, ensure that the column providing date values is sufficiently

detailed to calculate time state measures.

For example, if you define a time state measure called Inventory Level and your time dimension

contains Year, Month, and Week, ensure that the dates in your source data files provide details

down to the Week level in the time dimension. You can set the level of detail for the column asso-

ciated with your time dimension using the Degree of Detail property on the Column property sheet

(Time tab).

Message TR2731Cube group <name> is not defined in the primary drilldown path in dimension<dimension_name>. A cube group must be specified in the primary drilldown of adimension that has alternate drilldowns.

You cannot define cube groups based on a target level that exists in an alternate drill down. Cube

groups can only exist for levels in the primary drill down.

To remedy this problem, you can use the Drill Category property sheet (General tab) to make the

drill-down path in which you are defining the cube group the primary drill down. However, before

you do so, ensure that the current primary drill down does not contain specifications (such as

allocated measures and partitioning) that are not allowed in alternate drill downs.

Message TR2732The target category < > in cube group < > which summarizes external categoriescannot be suppressed or cloaked in any dimension views in dimension < >.

When you define a cube group, you cannot apply a dimension view that omits one of the categories

in the target level.



For example, suppose you create a cube group based on the level State (in the Regions dimension),

and the level contains three categories: California, New York, and Massachusetts. You cannot then

apply a dimension view that suppresses or cloaks any of those three states.

Message TR2733Data Source <data_source_name> is not associated with any dimensions that areincluded in cube<cube_name>. This data source will be ignored during .mdc generation.

In the Dimensions tab (PowerCube property sheet), you have omitted one or more dimensions with

which the named data source is associated. Because the values in the dimension are not required

for the PowerCube, Transformer ignores the data source when it creates the .MDC file.

Message TR2734All cubes have been disabled. MDC generation is not possible.

Check the PowerCube Status dialog box (Tools menu). All the PowerCubes defined for the model

have been set to a state of Disabled. You must enable at least one PowerCube in order for Trans-

former to create PowerCubes.

Message TR2735Measure < > has an average regular rollup with a weight measure < > specified. Bothmeasures must exist in the same data source in order for the weight measure to beapplied.

Ensure that measures used to weight other measures exist in the same data source.

Transformer analyses one data source at a time, and so both measures must exist in the same data

source.

Message TR2736Measure < measure_name1> has an average time state rollup with a weight measure< measure_name2> specified. Both measures must exist in the same data source inorder for the weight measure to be applied.



source.



Message TR2737Measure < measure_name1> has an average duplicates rollup with a weight measure< measure_name2> specified. Both measures must exist in the same data source inorder for the weight measure to be applied.



source.

Message TR2740Measure <> has a time state rollup specified. There is no time dimension in this model,the time state rollup will have no effect.

Check that the rollups set for the measure are defined correctly. You may have inadvertently set a

time state rollup instead of a regular rollup. Time state rollups have no effect on models that do

not contain a time dimension.

Message TR2800Transformer couldn't read the connection information for database <database_name>from file COGNOS.INI.

Transformer is unable to read the database information for a database that you are accessing via

an Impromptu IQD file. Check the entry for the database in the Cognos.ini or COGCONNECT.ini

file.

Message TR2801Transformer couldn't initialize the connection. Check the PowerGrid .DLL files in yourinstallation.

Transformer cannot communicate with the server.

Check your communications .dll's. Look for the file mwskj.dll in your Cogapps directory. You can

find out where your Cogapps directory is located by looking in your Cognos.ini or COGCON-

NECT.ini file.

You may be using command line options incorrectly. This error can be caused by running either of

the following command lines:

● rsserver -w ...

● rsserver.sh ...

The -w switch is not a valid switch for initiating rsserver in the command line. (This switch is used

by netd to start rsserver in client-server mode.)



Running the rsserver.sh script will also attempt to start rsserver in client-server mode, because the

final line contains rsserver -w.

Some valid command line formats are:

● rsserver -c -p model.py?

● rsserver -c -m model.py?

Message TR2802Transformer couldn't start the remote server task.

Transformer is unable to start rsserver, the server version of Transformer. Verify the following:

The owner of the rsserver files has not been changed from root to another user after installation.

The entry in the COGCONNECT.ini file matches the following:

[ Service - Transformer Server]

network=rsserver.sh

The -w option has not been removed from the line in the rsserver.sh file that launches server

Transformer. For example,

# launch Transformer in PowerGrid mode...

exec $COGNOS_HOME/rsserver -w

If the -w option is removed from that line, Transformer will not launch in PowerGrid mode.

The PowerGrid network daemon (netd) is running on the server and that you can communicate

with it. Use NetInfo to test your connection. You can use one of the following commands to

determine if PowerGrid has been started:

CommandOperating system

ps -ef | grep netd | grep cognosAIX

ps -ef | grep netdHP-UX

ps -ef | grep netd | grep -v grepSun Solaris

The shell script registered in COGCONNECT.ini (normally rsserver.sh) exists on the server and is

named correctly.

The first line of rsserver.sh or the netpgrc file points to the location of the .sh executable (Bourne

shell). By default, it should read something like #!/usr/bin/sh or #!/bin/sh.

The syntax for setting the database's environment variables in rsserver.sh is correct. If a shell script

is being launched from rsserver.sh to set environment variables for the database, ensure that the

script runs without errors.



Message TR2803Transformer couldn't connect to the active server task.

Transformer is unable to connect to a running server either because it is busy or no longer running,

or because the model has become corrupted.

Message TR2804The client/server connection has failed.

Transformer is unable to communicate with Server Transformer. There are several possible causes

of this error.

Ensure that the PowerGrid network daemon (netd) is active on the server. You may need to bring

netd down and restart it.

Ensure that your communications port setting is set to match the one defined for PowerGrid on the

server.

The problem may be that fin_wait_2 may be running; it conflicts with netd. Fin_wait_2 is an HP

process that runs on UNIX machines. The process is associated with the port number for netd. To

determine if it is running, check the processes running on the UNIX machine that are associated

with the port that is used by netd:

netstat -n |grep 1526 or WAIT

The only way to get rid of fin_wait_2 is to reboot the system. To resolve the problem, start netd

under a different port ID. Ensure that the new port ID is also reflected in the /etc/services file and

in Netinfo on the client. HP has released a patch to resolve this problem; it kills the fin_wait_2

process after a specified amount of time.

This error may also occur when a new version of Transformer is installed on the client, but the

upgrade for rsserver has not been installed. The version of Transformer on the client should be

accessing the same version of rsserver on the server. If there are currently two versions of rsserver

active on the system (for example, Transformer 5.21 and 6.0), then it is necessary to have two Netd

processes running on separate port numbers. Contact your UNIX administrator for further assistance.

The problem may be insufficient disk space in the directory on the UNIX system to which temporary

files are written. Try to increase the disk space. Otherwise, set the following environment variables

(either in rsserver.sh or in the trnsfrmr.rc preferences file) to point to a location that has sufficient

disk space:

DataWorkDirectory=<UNIX directory> ; export DataWorkDirectory ModelWorkDirectory=<UNIX

directory> ; export ModelWorkDirectory

Message TR3006Access Manager detected an invalid parameter value. Call Cognos Support.

This error message indicates that auto-access permissions are enabled but Access Manager does not

have valid cubes set up for auto-access.



A workaround is to move the Access Manager file to a different directory. Errors may appear but

the model will be accessible. Clear the Include Access Manager Auto-Accesses in the Model box in

Model Properties (Authentication tab).

Message TR3010Access Manager configuration < > could not be found or contains invalid information.

An Access Manager authentication file couldn't be found. When you specify an Access Manager

file to be used as an authentication source in a model, the name of the file should not include its

path. It should only be the name as it is defined in the Access Manager Administration utility.

Message TR3014The file system has denied access to Access Manager.

The authentication file may be in a read-only directory or on a read-only drive. Transformer must

be able to write to the authentication file. Remove read-only restrictions from the authentication

file.

Message TR3101Transformer couldn't create the directory < >.

You may have insufficient space in your temp directory or on the drive where the temp directory

resides. Try the following:

● Free up some disk space.

● Check free conventional memory and swap file size.

● Run Scandisk and Defragmenter to ensure that the free hard drive space is contiguous.

Message TR3311An error occurred during data retrieval from the database. Refer to Details for moreinformation.

If you are attempting to create a PowerCube using IQDs which access an Oracle data source, and

this error occurs only during peak network traffic times, the problem may be that the expire_time

parameter has been changed in the sqlnet.ora file. The expire_time parameter has a default of zero.

Contact Oracle for more information about this file and parameter, and do not modify the setting

without the express permission of the Oracle database administrator.



Message TR3439The temporary checkpoint file has become corrupted. Restart Transformer and reopenthe model file.

For some reason, the checkpoint file (.qy?) changed unexpectedly or was deleted while in use.

To solve this problem, save your work, close Transformer, then reopen your model file in Trans-

former.

Message TR3505Cannot open the .dbf file < >.

There is a problem with the .dbf file that is used as a data source.

Ensure that the .dbf file name is eight characters or less, and check that the path and name are

correct.

There may also be a problem with the Type setting. The file type is set in the Source type box, found

on the Source tab of the Data Source property sheet. Or in MDL, it is set with the SourceType

option in the Data Source definition statement.

Message TR3619Can't open the spreadsheet file < >.

The most common cause of this error is that there is a conflict between the two locations in

Transformer where paths are set. Paths for source files are set in two locations.

The first is the Data Source box in the Directories tab of the Preferences dialog box. Preferences

are also located in the file Trnsfrmr.ini.

The other is the Local Data File box on the Source tab of the relevant Data Source property sheet.

Or in MDL, this local data source is set with the option Source in the data source definition state-

ment.

Check that there is no conflict between the two settings. For example, if the Local Data Source

setting is a relative path, it will attach to the Data Source path and the resulting path must be valid.



Glossary

.iqd file

See Impromptu Query Definition File (.iqd).

.mdl file

A file that represents a model and is stored in plain text format using Transformer's MDL (Model

Definition Language). MDL files can be opened in Transformer or in any text editor.

.ppr file

A file format that can be read by PowerPlay Windows client.

.py? file

A model file stored in the default binary format. The question mark (?) in the extension .py? is

replaced by the character used in your version of Transformer.

See also, MDL File.

.qy? file

A temporary file that contains checkpoint entries of each major stage in the cube creation process.

Because this file is deleted when the process ends normally, the existence of a .qy? file indicates that

Transformer terminated unexpectedly. The question mark (?) in the extension .qy? is replaced by

the character used in your version of Transformer.

activate

Open an OLE object in its source application so that you can explore or change the object. If an

object is activated in place, it opens in the window in which it is linked or embedded, and the

toolbars and menus of the source application merge with those in the window.

advanced subset definition

A query that selects a new set of categories based on a dimension, drill-down path, and category

level or levels. You can further restrict the selection by adding find-in-cube and value restrictions,

and by including or excluding parent categories.

See also Find-In-Cube Subset Definition, Parentage Subset Definition, Value Restriction, and Subset

Definition.

alias

A different name for a table or entity so that the table or entity appears more than once. You use

aliases to resolve self-joins between tables, or to resolve ambiguous query paths by associating aliases

with different relationships than those of the original entities.

Alias is an Impromptu term that is called synonym (for tables in the data access layer).

See also supertype entity.


allocation

The process of placing summary data in a level, dimension, or category. Values can be allocated as

constants or in proportion to another measure.

alternate drill-down path

One of the drill-down paths in a dimension that contains multiple drill-down paths. Some tasks,

such as partitioning and allocating, cannot be performed on alternate drill-down paths.

In IBM Cognos Visualizer, alternate drill-down paths are identified by italics in the user interface.

ancestor category

In PowerPlay, a category above another category along a drill-down path. For example, 2003 is an

ancestor category of 2003/Apr.

apex

Omits the ancestors and siblings of a category from the dimension view. The base category in the

resulting dimension view is the apexed category. When a cube is created from this dimension or

user class view, users see only the apexed category, its descendants, and any special categories that

reference them.

An apex action can be performed only in view diagrams.

See also, Exclude.

association

A link between two objects.

In Transformer, the link between an item in a Transformer model, such as a level, and related

columns, attributes, or tables in the data source. An association specifies a type, such as a column,

and a role, such as a source, label, description, or drill-through target.

In IBM Cognos Visualizer, the link between a filter and the chart, panel, or scene that uses the same

data source. A filter associated with a data source is automatically applied to new charts that use

the same data source. The author who created the association can remove the association anytime

using the filter property sheet.

attribute

A property that qualifies, identifies, classifies, or quantifies an entity. For example, a Customer

entity might contain the attributes: Customer Name, Customer Code, and Address.

AutoDesign

A feature of Transformer that generates a preliminary model of dimensions and measures based on

the relationships it detects in source data.


Glossary

automatic exceptions

Exceptional values that are automatically highlighted. A value is considered exceptional if it deviates

significantly from the expected value computed from its row and column percentages.

autoregression

A method of forecasting based on the auto-correlation approach to time series forecasting.

Autoregression forecasting is most reliable when the driving factors of your business affect your

measures in a seasonal fashion.

bar chart

A chart that compares one item to another item or compares one item over time. A bar chart can

reveal trends and patterns.

In Visualizer the bar chart also reveals trends and patterns but also shows changes across a parameter,

such as a region or time, or compares two or more variables. You can create bar charts in two-

dimensional or three-dimensional format.

Bar charts can be sorted in ascending or descending order. Use a sorted bar chart to determine the

most or least significant contributors to a situation.

A progressive bar chart shows cumulative values. Each bar representing a measure value starts

where the previous bar ended.

See also chart, cluster bar chart, and stacked bar chart.

base value

The category against which proportions are calculated in Percent of Base calculations.

blanks

Blank rows or columns inserted in a report for visual effect. For example, you can use a blank row

and column to separate the data from the summary row and column.

calculated category

A category that shows the results of a calculation. Calculated categories are added to the cube in

Transformer. In PowerPlay Web Explorer, you can choose to hide calculated categories.

calculated column

A column whose values are calculated from other columns, functions, constants, and other calculated

columns.

calculated measure

A measure whose values are calculated from other measures, mathematical operators, and numeric

constants in an arithmetic equation.

See also measure.


Glossary

calculation precedence

The order in which PowerPlay for Windows evaluates intersecting calculations. The calculations

are performed in the following order:

1. calculated measure (after rollup calculation)

2. calculated category

3. dynamic rollup

4. regular rollup (with optional currency conversion calculated before rollup).

category

A division within a system of classification.

In PowerPlay, a category contains items in the rows, columns, and layers of a report

In Transformer and in IBM Cognos Visualizer, categories can be different levels of information

within a dimension. Categories can be grouped into more general categories. For example, a set of

dates can be grouped into a month, and months into quarters, and quarters into years.

In IBM Cognos Finance, a group of related reports, input forms, or libraries. Two levels of categories

are used to organize libraries and forms: first-level categories lead to second-level categories that

lead to a list of items. This branching is called the category path.

category code

A name that uniquely identifies a category within a dimension. The category code is generated by

Transformer when it creates the category.

See also, category, category label, object ientifier, source value, and special category.

category count

A measure that records the number of unique, non-zero, non-missing values for the categories in

the dimension and level you specify.

category label

A name that identifies a category.

See also category.

category set

A subset of the categories in a dimension. The categories can be from a single level or from different

levels in the same dimension. When you create a calculated category in Transformer, some functions

allow a category set to be specified as a parameter. Category sets are a convenient way of grouping

categories when you don't want to include the entire level in your formula.

category viewer

The portion of the dimension diagram that shows the category hierarchy of the selected dimension

and provides a mechanism for manipulating the categories.


Glossary

child category

A category one level below another category in a drill-down path. For example, 2000/Q1 is a child

category of the parent category 2000.

The child value is usually a refinement of the parent value. For example, 2000 Q1 is the child of

2000.

children

Categories one level below another category along a drill-down path. For example, Outdoor

Products, Environmental Line, and GO Sport Line are children of Products.

cloak

To remove a and its descendants from a dimension view, but summarize the values in the ancestor

category. The category is not generated in any cube that uses the dimension view, and it is not

viewable in any cube that uses the user class view. The category remains in the model. This option

combines the features of summarize and suppress. You cannot cloak special categories.

clustered bar chart

A chart that groups related information, compares summaries, and compares categories or measures

across time or any other variable. You can create a measure-based or a dimension-based cluster

bar chart. Clustered bar charts can be either two-dimensional or three-dimensional in type.

See also bar chart, measure-based chart, and dimension-based chart.

cognos security administration file

A .csa file is used by Access Manager to store connection information. It maintains directoryserver

and .lae file connection information, the directory server currently configured to be active,and the

expansion state of the nodes on the directory tree. The Access Manager Configuration wizard can

use a .csa file to automatically set runtime configuration information.

collection

A group of related OLE objects that you can reference as a unit. Any action performed on a collection

affects all objects in that collection.

See also ownership collection and reference collection.

column

One kind of information in a list frame, it is a vertical list of data that shares the same definition.

condition

An expression that can be used anywhere you need a true-false value. You can use conditions in

filters, and in calculations. For example, you can use a condition to see only data for your sales

region.


Glossary

consolidation

The process of combining two or more records from the data source into a single record in the

cube. Consolidation is possible only for records that have the same set of dimension leaf values.

Measure values from the contributing records are combined by applying the rollup function rules

specified for the measures.

constant

Represents a numeric or string value that doesn't change. True, False, and Null are examples of

constants.

control cube

A cube that contains the structural information used to combine multiple time-segmented PowerCubes

into a time-based partitioned cube.

convergence level

The level at which two or more alternate drill-down paths meet.

correlation display

A graph that shows the values of two measures that are being compared. Bars represent one measure

and a line represents the other. You must have at least two measures to use a correlation display.

crosstab chart

A chart that uses a row and column format to show measures for different combinations of categories.

For example, you have a crosstab chart with Product categories as columns and Years categories

as rows. The cells contain the value for the Total Revenue measure for every combination of Years

and Product categories.

cube

A multidimensional object used by many IBM Cognos products to retrieve data at any level of

aggregation, across any set of dimensions, for reports or plans. You can store cubes in supported

formats in a LAN folder, or on a local computer.

A multidimensional representation of data. A cube contains information organized into dimensions

to provide faster retrieval and drill down in reports.

PowerCubes are created in PowerPlay Transformer or PowerPlay Connect. PowerPlay Connect is

distributed with IBM Cognos Visualizer. Authors can use PowerPlay Connect to create or modify

pointer files. A pointer file is an .mdc file that contains connection information for an IBM Cognos

data source, or an OLAP server.

cube group

A set of similar cubes built by Transformer. Each cube group relates to a single level in one

dimension of the model. Each member of the group is targeted at one of the categories in the level.


Glossary

cube object

An object in a model that corresponds to a cube. Specifies how to build the cube and how to reflect

the status once the cube has been built.

currency record

The format information that PowerPlay uses to display the currency data. The currency information

is stored at the same level of detail as the date dimension of your model.

currency table

A table of information about currencies that you maintain in Transformer or load from an external

data source A base currency table shows the default currency of the country specified in the Locale

setting of your operating system, such as U.S. dollars. A Euro table contains the fixed conversion

rates used to convert values to or from the national currencies of EMU countries using the Euro as

the intermediate currency. You need both a base table and a Euro table for conversions between

non-EMU and EMU currencies after 1998.

current period

A category in a time dimension that Transformer uses to determine the composition of relative time

categories in the dimension, if any. Each time dimension has its own current period.

For example, if a time dimension contains a relative time category for Last Month, and the current

period is set to February 2006, Transformer determines that Last Month is January 2006. When

the current period in the dimension changes, Transformer updates the Last Month category.

custom exceptions

Exceptional values that are defined as rules and applied to categories in a report. A value is considered

exceptional if it falls within the defined value range.

custom time dimension

A dimension that derives its values from more than one column and whose structure is created

manually.

See also, time dimension.

data item

In IBM Cognos Visualizer, a data item is a text or numeric component of a chart or filter, which

corresponds to a dimension, a measure, a column, or a category within a dimension.

See also element.

data source

The originator of the computer data that your product works with. For example, a data source for

an IBM Cognos product may be a relational database or a file that uses one of the supported data

formats.


Glossary

data source directory

The directory in which Transformer searches for source files used in a model. You can set the data

source directory in the Directories tab (Preferences dialog box).

data source scope map

A color-coded dimension map that shows the association of dimensions and levels with each data

source and measure in the model.

See also, measure scope map.

date_exp

A date constant, a date data item or any expression resulting in a date value.

datetime_exp

A date constant, a date data item or any expression resulting in a datetime value.

delimited field text file

A file containing text data, where the fields in each record are separated (delimited) by a character,

such as a comma or a tab.

descendant category

Any category below another category along a drill-down path. For example, 2003/Apr is a descendant

of 2003.

See also ancestor category and child category.

diagram

A graphical representation of categories in all or part of a dimension and the relationship of the

categories to each other.

See also, dimension diagram, dimension view, and view.

dimension

Descriptive data about a major aspect of a business, such as products, cost centres, or markets,

upon which reports and plans are based. Each dimension can include different levels.

dimension diagram

A diagram that shows an entire dimension structure, including the primary and alternate drill-down

paths, and special category structures.

dimension line

In PowerPlay for Windows and PowerPlay for Excel, one or more rows that shows the categories

from each dimension used to filter on data for the current report.

In PowerPlay Web, one or more rows that shows the categories used to filter the data from each

dimension in the cube. The dimension line appears above the display in your Web browser.


Glossary

In Transformer, the row of dimension names that appears either along the top of the dimension

map window or just below the Measures and Data Sources option buttons in the Show Scope

window.

dimension map

A table that shows the Transformer model in rows and columns. The columns in a dimension map

are the dimensions, for example, Dates, Products, Sales Regions. The rows in a dimension map are

the levels within the dimensions, for example, year, month, day (for a time dimension) or east, west,

north, south (for a regions dimension).

dimension menu

In PowerPlay for Windows and PowerPlay for Excel, a menu that appears when you select a

dimension folder. You use a dimension menu to filter on specific information. When you select a

category from the menu, the level of the dimension changes and so do the values in the report.

dimension table

In star schema designs, a data source that contains columns that map to levels and categories to

build a dimension in a Transformer model. Also referred to as a structural data source because it

defines the structure of the model. A dimension table differs from a fact table or transactional data

source because it does not include measure values.

For more information, see the PowerPlay Transformer documentation.

dimension view

A subset of a dimension that is used to create cubes that contain only selected aspects of the data

represented by the complete model. For example, views based on specific countries or regions are

a useful way of ensuring that users see only the data that is most relevant to them.

dimension viewer

An organized view of all dimensions, levels, and categories in the selected cube. Use the dimension

viewer to add categories as rows or columns and to filter information.

You can also use the dimension viewer in Windows to add categories as layers, create subset

definitions, format measures, and define sets of categories.

directory server

A general term for an LDAP-compliant server that contains authentication data. IBM Cognos

applications can use the SunONE directory server or Active Directory Server to associate users with

data access permissions.

display

A type of chart or graph.

You can change the display by clicking a display button on the toolbar.


Glossary

drill

An action that shows different categories. Drilling down shows child categories. Drilling up removes

child categories from the report.

drill category

The immediate descendant (child) of a root category. Used only to define the properties of a drill-

down path.

drill down

To follow a link from one layer of data to a more detailed layer. The layers are set by the structure

of the data. This action shows child categories.

See also drill up.

drill down path

The successive root category and leaf categories available from a drill category.

drill through

To view the information linked to a value in a report, cube, or macro, or an Impromptu report,

Impromptu Web report, PowerPlay cube, PowerPlay report, or PowerPlay Web report. For example,

you can drill through a value to view the detailed sales transactions for a particular customer. Any

filtering of information in the original object is automatically applied.

drill up

To follow a link from one layer of data to a less detailed layer. The layers are set by the structure

of the data. This action shows parent categories.

See also drill down.

driving category

A category whose values are compared with those specified in the custom exception definition.

duplicate record

A record with categories that are identical in all dimensions. For example, the following source

data records are duplicates, because they all represent the state of New York in the Northeast

Region of the USA, on April 13, 1999:

19990413, USA, NORTHEAST REGION, NEW YORK, 1.0, 100.00,200.00



embed

Inserts information created in one application into another application. Once embedded the

information, called an object, becomes part of the document in which it is embedded. When you


Glossary

double-click an embedded object, you open the application in which the object was created where

you can edit the object.

You can embed objects in an Impromptu report or you can embed Impromptu report into another

application.

See also link and OLE (Object Linking and Embedding).

exception highlighting

Formatting that is applied when the information in the report meets the conditions set by the

automatic exception sensitivity, or by the application of a custom exception.

exclude

In Transformer, to omit a category and all data associated with the category and its descendants

from a dimension view or user class view. The category and its descendants are not generated in

any cube that uses the dimension view. The category remains in the model.

In Transformer versions prior to 6.6, this command was called filter.

explain window

A window that shows descriptive information about the selected object.

In Transformer, the administrator provides descriptions of dimensions and categories when creating

a cube.

In IBM Cognos Visualizer, the author provides descriptions of charts, panels, scenes, and sheets.

Explorer report

A report that shows one level of one dimension in each row and column. You can add more levels

from the same dimensions or from different dimensions as nested categories.

See also, Reporter report.

exported model file

See MDL File.

expression

Any combination of operators, constants, functions, data values, and other components that evaluates

to a single value.

fact table

In star schema designs, a data source that contains records which provide measure values for cubes.

Also referred to as a transactional data source. In combination with one or more dimension tables

or structural data sources, fact tables populate the model in Transformer.


Glossary

feelers

A line connecting a chart element to a chart label. In IBM Cognos Visualizer, feelers are available

on the pie, gauge, box and whisker, and thermometer charts. Use feelers when chart labels are in

close proximity, making it difficult to find the corresponding chart element.

filter

A set of criteria used to retrieve a subset of records. For example, instead of viewing total sales,

you can view sales for a specific region, product line, or time period.

You can add filters to entities, use filters to create specialized subtype entities, and use filters to

ensure consistency in the filter expressions used by report users in Impromptu, PowerPlay, and

Transformer. You can also restrict the data that users can see by applying filters to objects within

a user class. For example, you can create a filter on the Salary table so that only the Human Resources

department can view that table.

filter bar

In Microsoft Excel, a toolbar that appears when the PowerPlay for Excel add-in is activated. The

filter bar shows the dimension folders and is used to filter on specific categories or measures.

find-in-cube subset definition

Subset definition created using the Find in Cube tool, which searches for the text you type in the

Find What box. You can find all categories or dimensions in a cube, which meet specified criteria.

You can have a case-sensitive search, a whole category search, or a wildcard search.

See also Advanced Subset Definition, Value Restriction, and Parentage Subset Definition.

Find in Report

A feature that searches for data in the report using pattern matching and wildcards. You can search

any or all layers, and search row, column, or layer labels.

fixed-field text file

A file that contains text data aligned in fixed-length columns.

forecast

A calculation of future value by using existing values on the current display. In PowerPlay, you can

use forecast to predict future values beyond that currently appearing in your window.

function

A function is a subroutine that returns a single value. You can use functions to create calculations

and conditions to filter data. Functions are similar to operators in that they manipulate data items

and return a result. Functions differ from operators in the format in which they appear with their

arguments. This format allows them to operate in zero, one, two, or more arguments:

function (argument, argument, ...)

Functions are of these general types: Date and Time, Numeric, Summary, and Text.

For more information about using functions, see the Expression Editor online help.


Glossary

growth

A method of forecasting based on the exponential regression technique of time series forecasting.

Growth forecasting is most reliable when the driving factors of your business affect your measures

exponentially.

horizontal axis

Normally, the category or x-axis on a chart or graph. However, if the chart is re-orientated, the

horizontal axis may be the value or measure axis.

IBM CognosScript language

The IBM CognosScript language is a BASIC-like macro language provided with all IBM Cognos

Business Intelligence Products.

IBM Cognos security administration files (.csa)

A .csa file is used by Access Manager to store connection information. It maintains directory server

and .lae file connection information, the directory server currently configured to be active, and the

expansion state of the nodes on the directory tree.

Impromptu Query Definition

A file (.iqd) created by Impromptu that defines a relational database query. You can use Impromptu

query definition files as data sources for visualizations and can define security for them.

incremental update

To rebuild a cube by refreshing a limited portion of the data. Using incremental update, you can

add new data to cubes without having to re-create the cube from scratch. Older data is maintained.

indented layout

A layout for crosstabs. In the rows area, categories from different levels are laid out vertically, with

indentation.

See also Standard Layout.

integer_exp

An integer constant, an integer data item or any expression (including a numeric_exp) resulting in

an integer value.

intersected categories

A feature that combines categories from different dimensions. For example, you can select Products

and Locations, and add the new intersected category as rows. If you add Years as columns, the

report shows the number of products sold over the years, by region.

This feature is only available in Reporter reports.

interval_exp

An interval constant, an interval data item or any expression resulting in an interval value.


Glossary

label

A label identifies a position in the program at which to continue execution, usually as a result of

executing a GoTo statement. To be recognized as a label, a name must begin in the first column,

and must be immediately followed by a colon (":"). Reserved words are not valid labels.

layer

A set of categories for a dimension. You can add a layer to a report to provide a new perspective

on your report results. You can look at only one layer at a time. For example, a report shows the

number of tents sold in California between 2000 and 2001. You add another layer to show the

value of the tents sold. Both layers show the same rows and columns, but they use different measures.

leaf category

A category at the lowest level of detail in a dimension. For example, if a dimension contains the

levels State, City, and Store, and has within these levels the categories California, San Diego, and

Pro Form Supplies, then Pro Form Supplies is a leaf category.

legend

An explanatory list of categories in the report for certain displays. It shows the category name and

color representing the associated data. The legend doesn't appear in crosstab, simple bar, simple

line, and 3D bar displays.

level

A position in a hierarchy of data where all members have common or default attributes. When you

drill up or down, you change levels. For example, when you drill down on years, you may go to

quarters. Levels can come from the data source (source levels).

level heading

The top part of a diagram, which shows the name of a level in a dimension.

line chart

A chart that tracks one or more variables, compares trends and cycles, or shows a time series analysis.

You can create single or multi-line charts.

link

The connection between a linked object and the application in which it was created. When the

information changes in the source file, the changes are reflected in the destination file. You can

choose to update linked information manually or automatically.

See also embed and OLE (Object Linking and Embedding).

list

A window that shows the names of all data sources, measures, cube objects, or user classes in a

model.


Glossary

local authentication

The process of verifying access to protected data sources using local authentication export files

(.lae) or local authentication cache files (.lac) Usually used for mobile or standalone users.

local authentication export file (.lae)

A source of authentication data in Access Manager that is independent of a directory server and

that may be used to

● authenticate standalone users who cannot be authenticated over a network (Access Manager

configuration)

● transfer authentication data between namespaces (Access Manager Administration)

You can only use .lae files locally on a single computer, not on a network or with multiple users.

locked dimension

A dimension to which new categories cannot be added. Records that belong in new categories are

discarded. Locking a dimension locks all the levels within it. A lock icon appears in each level.

Levels can be locked individually.

See also, Locked Level.

locked level

A level to which new categories cannot be added during category generation. Records that belong

in new categories are discarded. A lock icon appears in each level. A dimension can also be locked.

See also, Locked Dimension.

LOG File

A record of the activities that take place during a computer process. For example, a log file contains

information about model processing in Transformer, or server activity in PowerPlay Enterprise

Server.

long name

The comprehensive or complete name for a dimension or category, which is often accompanied by

a shorter, abbreviated name. For example, the long name could be All Products and the short name

Prods. Long names are defined in Transformer.

See also Short Name.

lowest detail

The lowest level of information available. For example, you have a category called 2006 that is

divided into four quarters. Quarters are divided into months, which in turn are divided into weeks.

The lowest level of detail available for that dimension is weeks.

See also Leaf Category.


Glossary

macro

A customized sequence of instructions ("macro commands") that IBM Cognos' applications can

carry out.

manual level

A level that has no associated source column. Categories in manual levels must be created and

maintained manually. In the dimension map, manual levels appear with a hand icon.

markers

The graphical representation of a single data point in a display. Markers can be found in correlation,

multiline, scatter, surface plot chart and single line displays.

MDL (Model Definition Language)

A proprietary language that duplicates the functionality of the Transformer user interface. MDL is

compatible between different versions of Transformer. Transformer model files that are formatted

for export use the .mdl extension.

measure

A performance indicator that is a quantifiable set of values and gives information about the per-

formance of the business. Examples of measures are revenue, revenue per employee, and profit

margin percentage.

In IBM Cognos Visualizer, measures refer specifically to numeric data items that come from multi-

dimensional data sources, whereas numbers refer to numeric data items in tables or flat files

measure folder

A folder that groups measures from a PowerPlay model into logical groupings. A measure folder

can be hierarchical and group either like measures to be collected or calculated measures based on

existing measures.

In IBM Cognos Visualizer, calculated measure folders can be referenced in the creation of a chart

or filter, whereas measure folders that merely group like measures cannot.

See calculated measure.

measure scope map

A color-coded dimension map that shows the association of dimensions and levels with a measure.

metadata

Descriptive data about the data contained in a database. For example, the property sheet for a

column contains metadata about that column, such as name, data type, and size.

mirror reference

In PowerPlay Enterprise Server, a mapped cube or report that forwards incoming requests to be

processed on a computer other than the one where the object is added.


Glossary

missing value

A value that is not available for a category because of the context, or because there is no data.

Missing values appear in a report as zeros, NA (not available), a blank (nothing in the cell), or

missing, depending on how the measures in the cube were designed to handle missing values.

model

See business model and Transformer model.

multiline chart

A chart that reveals and compares trends and cycles that show relationships between variables. It

also shows time series analysis and relationships between variables.

namespace

A source of authentication data used by Access Manager that exists as a directory on a directory

server, or as an entry in a local authentication export file (.lae), depending on the default security

server configured in the system registry.

The security data stored in each namespace, such as signon information for users, user classes,

application servers and data sources, distinguishes each entry from all other namespaces in the

repository.

nested category

Categories are arranged in multiple levels along the rows, columns, or layers in a crosstab display.

Nested categories form groups of information that add another perspective to a report.

nested chart

A graphical display of nested data. You can view multiple charts representing the outer rows and

columns in one display. In PowerPlay for Windows, this option is not available for horizontal bar

charts.

nested crosstab

A crosstab report that contains more than one data item in the rows or columns. For example, you

can nest Product Type and Product Line beside each other in the rows.

NetInfo

A utility, packaged with PowerPlay and installed locally, which validates connections to server-

based Transformer models and issues commands to PowerGrid.

NewsBox

A folder that can contain other NewsBoxes, shortcuts, reports, agents, custom views, URLs, tools,

and data sources.

NewsItem

An object stored in the Upfront data store that represents a report, agent, custom view, data source,

tool, or URL.


Glossary

numeric_exp

A numeric constant, a numeric data item or any expression resulting in a numeric value.

object identifier

In Transformer, a numeric value, unique within the model, that identifies a category. It appears in

the title bar of the Category property sheet if you select the Object Identifier check box from the

Titles tab of the Preferences dialog box. Transformer generates the ID when it creates the category.

See also, Category Code, Category Label, Source Value, and Special Category.

OLE (Object Linking and Embedding)

A Microsoft Windows facility that enables applications to share data. When you use OLE, two

applications can share data through a connection that you establish.

See also link and embed.

operator

Specifies what happens to the values on either side of the operator.

There are four types of operators:

● Arithmetic: performs arithmetic operations on two parts of an expression

● Comparison: compares one or more values

● Logical: defines relationships between two parts of the expression

● String: concatenates two character strings

orphanage

A manually-created category in a manual level that acts as a parent category for newly generated

categories. The orphanage category first must be designated in the dimension diagram.

See also, Source Orphanage.

overflow value

A value that is larger than the measure's storage type allows. An overflow value in a report can be

erroneous or may return an error. The administrator defines the measure's storage type when cre-

ating the cube.

parentage subset definition

A subset definition based on the parent/child relationships below a specific level. You can base the

subset definition on the categories at the next level down, at the next two levels down, or at the

lowest level of detail.

See also Find-In-Cube Subset Definition, Value Restriction, and Advanced Subset Definition.

parent category

A category one level above another category along a drill-up path. In PowerPlay, a category can

have more than one parent if it lies along more than one drill-down path. In IBM Cognos Visualizer,


Glossary

you can only follow one drill-up path. The parent value is usually a consolidation of all its children's

values. For example, 2001 is the parent category of 2001 Q1, 2001 Q2, 2001 Q3, and 2001 Q4.

partition

A nested sub-cube that Transformer can create to speed data access.

partitioning

In Transformer, a process that presummarizes portions of the cube data into nested sub-cubes called

partitions, resulting in faster data access for report creation.

PDF File

See Portable Document Format File (.pdf).

performance indicator

A numeric attribute of the business model that is quantifiable, and is identified as a performance

indicator of your business. When you specify an attribute as a performance indicator, it identifies

the attribute as a measure candidate for the Metadata Explorer in PowerPlay Transformer.

See also descriptive indicator and key performance indicator (KPI).

pie chart

A graph that shows the relationship between the whole and the parts. For example, in PowerPlay,

a pie display can show you how much of a department's budget goes to paper supplies. In IBM

Cognos Visualizer, you can create three-dimensional pie charts with a number axis that you can

show and scroll through.

Pointer File

An .mdc file that contains information about the actual data source. For third party cubes, a pointer

file stores the connection information.

Portable Document Format File (.pdf)

A file format that maintains the formatting of the original document. You can view and print a

.pdf file without the need for the authoring application or fonts.

PowerCube

A file that contains data that is structured to provide for fast retrieval and exploration in PowerPlay.

PowerCube group

See Cube Group.

PowerCube object

See Cube Object.

PowerGrid

Software used to communicate between the versions of Transformer on Windows and UNIX.


Glossary

PowerPlay Portable Report File (.ppx)

A file format that can be read by any PowerPlay client. You can create .ppx files in PowerPlay for

Windows and PowerPlay for Excel.

PowerPlay Web Explorer

A Web-based application that is used to explore multidimensional data. One of two applications

that make up PowerPlay Web.

PowerPlay Web Viewer

A Web-based application that is used for on-demand report viewing. One of two applications that

make up PowerPlay Web, PowerPlay Web Viewer renders PowerPlay Portable Reports in PDF

format.

PowerPlay Worksheet

In PowerPlay for Excel, a worksheet that is associated with a cube. When the Excel workbook is

saved, the query information required to reconnect to the cube is saved with the PowerPlay work-

sheet.

presummarization

Summarizing calculations that are performed when a cube is built.

See also, Summary Partition.

primary drill-down path

In PowerPlay for Excel, a worksheet that is associated with a cube. When the Excel workbook is

saved, the query information required to reconnect to the cube is saved with the PowerPlay work-

sheet.

Protected Cube

A cube where members of different user classes have access to specific categories, measures, or

dimensions depending on their access privileges.

query

A question to the database, snapshot, or HotFile that defines what data you want to retrieve from

the data source that can be executed repetitively.

Query Processor

The component of PowerPlay Enterprise Server that processes cube requests and generates results

for PowerPlay Web Explorer, PowerPlay for Windows, and PowerPlay for Excel.

rank

To assign an ordinal to values in the selected row or column. You can assign ordinals to the top or

bottom values, and sort the ordinals in ascending or descending order.


Glossary

rank category

A new row or column showing the rank ordinals.

regression

A statistical function that produces a curved line that is a best fit for logarithmic growth. A logar-

ithmic regression line can be added to a graph in PowerPlay for Windows. Regression is best used

when fitting the growth of a measure against a time dimension.

relative time category

A special category in a time dimension that allows cube users to create reports using time periods

such as Current Month, Last Month, and Quarter-to-Date, each of which is defined relative to the

current period.

Relative time categories are updated automatically by the administrator in Transformer, according

to the current period set for the dimension.

relative time function

A pre-defined function in a time dimension, defined by the properties of a relative time category,

that determines the related categories.

For example, a relative time category can use the Year-to-Date relative time function to generate

its related primary categories.

report

The presentation of information about a given topic prepared by using a layout defined by an

author. There are many types of IBM Cognos reports, including visualizations and Impromptu

requests.

In PowerPlay for Windows, a report contains the latest data selected from one or more cubes.

In PowerPlay for Excel, a report includes one or more PowerPlay worksheets and may include

additional Excel sheets.

In Impromptu, a report contains a view of the current data in your company database. What you

see depends on the data that you can access from your catalog.

In IBM Cognos Finance, a report is a database query that is structured like a spreadsheet. Unlike

spreadsheets, you can build and change the form of a report without affecting the database.

Reporter report

In PowerPlay for Windows, a flexible report that enables you to

● add single categories from different levels of the same dimension

● add categories from different dimensions

● add nested categories from the same dimension, or from different dimensions

See also, Explorer Report.


Glossary

Report Processor

The component of PowerPlay Enterprise Server that processes report requests and generates results

for PowerPlay Web Viewer.

role

A description of the association between an item in a model and a data source. For example, an

association may include a source role of product number and a label role of product name. In this

instance the category source value references product number and the category label references

product name.

See also, Association.

rollup calculation

An expression that is used to create a new calculated category that applies the rollup function (either

add, minimum, maximum, or average) specified by the measure for the selected categories.

rollup function

A method used to summarize measure values up the category tree structure. You can select rollup

function settings from the measure property sheet.

See also, Consolidation.

root category

The category in a dimension or subdimension from which all other categories are descendant.

row

Composed of one data value from each column in the database. Also known as a record.

scatter plot chart

A graph consisting of points whose coordinates represent values of data. A scatter chart/display is

used to compare two or three different measures. You can create either two-dimensional or

three-dimensional scatter charts.

You must have at least two number columns to create a scatter chart. You cannot use measures for

the x, y, or z axes in scatter charts because measures require at least one dimension. However, you

can use measures for the color and radius optional properties of a scatter chart.

See also chart.

shared dimensions

A common linked dimension that exists in two or more open reports. When you perform an action,

such as filtering, drilling, or slicing and dicing, the shared dimension in each report is updated. Any

combination of Reporter and Explorer reports can share the dimension line.


Glossary

short name

An abbreviation or code-name for a more comprehensive long name for a dimension or category.

For example, the short name could be Prods and the long name All Products. Short names are

defined in Transformer.

sibling category

Categories at the same level. For example, Outdoor Products, Environmental Products, and GO

Sports Line are sibling categories of the parent category Products.

slice and dice

Changes and arranges data when you

● choose different categories for your report

● drill down and drill up

● filter information

For example, if a report shows the number of products sold by each branch at the end of the last

quarter, you can slice and dice information to show revenue over the last two months for each

product line.

sort

Arranges values in numerical order or labels in alphabetical order. You can sort in ascending or

descending order.

source orphanage

A category created by Transformer to act as a parent category for new categories that don't have

clearly defined parent categories due to a lack of contextual information.

Later, if more information about the parent categories of the new categories becomes available,

Transformer automatically connects the categories to the appropriate parent categories.

By default, the suppress attribute of source orphanage categories is enabled.

source value

The data value retrieved from a source data record and used to identify or locate a category.

sparse data

A data source where the number of actual datapoints within a range is quite small compared to the

possible datapoints in the range.

In PowerPlay for Windows, the number of possible datapoints is the product of all leaf categories

in every dimension. This number is usually much larger than the actual number of data records in

a cube.

special category

A category that groups a set of regular categories from any level in the same dimension without

regard to their normal hierarchical organization.


Glossary

For example, you have a dimension called Management that includes the levels Senior Management,

Middle Management, and Junior Management. You can have a special category called Social

Committee that includes specific personnel from each of these levels.

spreadsheet crosstab

A spreadsheet that contains named ranges that become the columns in the model when imported

into Transformer.

See also, Spreadsheet Database.

spreadsheet database

A spreadsheet that contains a list of records in the form of a database table. You can use a spread-

sheet database as a data source for a model, provided the database contains a named range that

identifies the matrix of data to use.

See also, Spreadsheet Crosstab.

stacked bar chart

A chart that shows relative proportions of parts to the whole and the relationship between the parts

using stacked rectangles. You can create two-dimensional and three-dimensional stacked bar charts.

standard layout

A layout for crosstabs. In the rows area, categories from different levels are presented horizontally,

without indentation.

See also Indented Layout.

standard report

A saved report that updates when opened to reflect changes in the cube or cube data. The report

contains specific information such as subset definitions.

storage type

The format used to store values in the data source and in temporary work files. For example, source

values that represent text or alphanumeric category labels are stored as text.

stored procedure

A procedure defined and stored in a host database (such as Oracle or Sybase) that performs actions

on the database.

string

A character string enclosed in quotation marks. For example: 'IBM.'

string_exp

A string constant, a string data item or any expression resulting in a string value.

structural data source

See Dimension Table.


Glossary

style

A defined group of formatting options you can apply to selected attributes. When users run reports

that include that attribute, the corresponding report column is formatted using the information in

the style. A style can contain font information (such as type, size, and color), borders, patterns, and

data formatting.

You can define the following types of style: character, date, date time, local currency, number, time,

and time interval.

subdimension

A tree of categories with levels that are independent of levels in the dimension. A subdimension can

provide different details, or different levels of detail, for categories in a level. A subdimension can

support an alternate drill-down path.

subject

The representation of an entity in a package. You can specify query paths explicitly for each

attribute subject within a subject. Changes you make to a subject in a package do not affect the

underlying entity in the business model.

See also package and subject item.

subject item

The representation of an attribute, filter, or prompt in a package. Subject attributes contained in a

subject are based on the attributes contained in the entity the subject references. You can also add

subject items that reference attributes from other entities, as long as they are related unambiguously

to the main entity.

Changes you make to a subject item in a package do not affect the underlying object in the business

model.

See also subject and subject folder.

subset

A group of categories created to isolate information that shares some common criteria. Subsets are

created by adding a subset definition to a report.

See also Advanced Subset Definition, Find-In-Cube Subset Definition, Value Restriction, and Par-

entage Subset Definition.

subset calculation

A calculation that is performed on all members of a subset. Subset calculations include the following

calculations: add, multiply, minimum, maximum, or average. A subset calculation always appears

directly after the subset. The calculation is recalculated if the subset changes.

subset definition

Information that defines a new set of categories based on specified criteria saved with a report.

Subset definitions can be parentage subset definitions, find-in-cube subset definitions, and advanced

subset definitions.


Glossary

subset viewer

In PowerPlay for Windows, the pane in the bottom part of the dimension viewer. The subset viewer

shows subset definitions created by the parentage, find-in-cube, and advanced subset tools. You

can add individual categories from the results as a row, column, or layer, or you can use the resulting

categories to create a subset.

In PowerPlay for Excel, a window that shows subset definitions created by the parentage, find-in-

cube, and advanced subset tools. The subset view can be resized, closed, and undocked from the

Navigator window.

summarize

Sums all descendant categories for a category and suppresses the descendants.

See also, Cloak.

summary

A summary component combined with a data item. For example, Total (Sales).

summary partition

A partition that contains presummarized values for the categories in higher levels of one or more

dimensions. Information requests that can be satisfied from the summary partition use the presum-

marized values and, therefore, require less calculation at the time of the request.

suppress

Conceals a category within a dimension view. Within any cube that includes the view, the immediate

descendants of the suppressed category link directly to its immediate ancestor category. In the cube,

users see only the immediate ancestors and immediate descendants of the suppressed category.

See also, Cloak and Summarize.

swapping

To exchange the position of categories in a report. You can swap rows and columns, columns and

layers, or rows and layers. The terms rows and columns change depending on the current display.

For a pie display, rows and columns are called displays and slices.

synchronization

The process of transmitting information between a client and server model so that the models are

identical. Both model files must be in .py? format. The question mark (?) in the extension .py? is

replaced by the character that is used in your release of Transformer. Synchronization occurs every

time you connect to a server model from Transformer on Windows.

tick

Small lines that intersect an axis like divisions on a ruler.


Glossary

time_exp

A time constant, a time data item or any expressiont resulting in a time value.

time array

A contiguous set of four or twelve data columns in a data source file that represent measure values

for either four consecutive quarters or twelve consecutive months in a year.

time-based partition cube

A cube that combines multiple time-segmented PowerCubes based on the structural information

in a control cube.

time dimension

A dimension that contains categories representing time periods. Transformer indicates a time

dimension by showing a clock icon to the left of the dimension name.

A time dimension may derive its categories from the values in a single column, or be customized to

derive its values from more than one column.

By default, a time dimension is built with year, quarter, and month levels.

time level

A level in a time dimension that derives categories for years, quarters, months, and days from dates

in the data source.

For example, from a date such as 20060113 in the data source, Transformer can generate time

categories such as 2006, 2006/Jan, 2006 Q1, and 2006/Jan/13.

time state measure

A measure that represents the state of a system at a specific point in time.

For example, a state measure might show the number of employees in a company at the start of

each month, or the balance of a bank account on a given day each week. This type of measure is

distinct from transaction measures such as the amount of a sale.

transactional data source

See Fact Table.

trend

A method of forecasting based on the linear regression technique of time series forecasting. Trend

forecasting is most reliable when the driving factors of your business affect your measures in a linear

fashion.

triangulation

The process of converting one currency to another through a third. This method is used to convert

one European Monetary Union currency to another, or a non-European Monetary Union currency

to a European Monetary Union currency.


Glossary

For example, to convert from French francs to Italian lire, the francs are first converted to Euro,

and then converted from euro to lire.

unbalanced diagram

A diagram representing dimensions that each contain subdimensions that are wholly or partially

different from the subdimensions in other dimensions.

For example, a diagram shows the dimensions 1999, 2000, and 2001. The 2001 dimension contains

subdimensions for quarters and, within quarters, for months. The 2000 dimension contains subdi-

mensions for quarters only. The 1999 dimension has no subdimensions.

uniqueness

A designation for a level that indicates that each category in that level can be identified by its source

value alone, without reference to its ancestors. You must specify that your data is unique when a

level is the convergence level for alternate drill-down paths or when the model contains multiple

data sources.

user class

A group of users who are assigned the same data access privileges, usually because they perform

similar functions in an organization. User classes can contain other user classes, and users can

belong to several user classes. User classes are related to each other in a tree structure. Each class

inherits the rights of its parent, which can be further restricted but not relaxed.

User classes are assigned in Access Manager.

user class view

In Transformer, the categories and measures that members of a specified user class are permitted

to see, typically a subset of the information contained in the entire PowerCube. The cube designer

can specify whether the values associated with omitted categories are rolled-up (summarized) or

removed from reports based on the cube.

In other IBM Cognos applications, the term is used more generally to signify access to a data source

or an authorized subset of the information in that source, based on user class membership.

Note: Not to be confused with the User Classes and Users View in the Administration tool of

Impromptu Web Reports, which is a hierarchical view of the User Classes folder and the Users

folder.

value restriction

An advanced subset definition created using the Restrict by Value dialog box. You can restrict the

selected measure by highest or lowest values, or a range of values.

See also Find-In-Cube Subset Definition, Parentage Subset Definition, and Advanced Subset Defin-

ition.

variable

An identifier that's used to store a value in an IBM CognosScript Language macro.


Glossary

vertical axis

Normally, the measure or y-axis on a two-dimensional chart. The z-axis on a three-dimensional

chart. However, if the chart is re-oriented, the vertical axis may be the category axis

view

A definition stored in a database's metadata that defines a 'virtual table' that does not exist until

referenced in an SQL statement, such as a select statement.

ymdinterval_exp

A year-month-day interval expression.


Glossary


Glossary

Index

Symbols.asc files

as data sources, 49

.crr files

setting up drill through, 124

.csv files

as data sources, 49

.imr files

drill through from PowerPlay, 124, 126, 127

.ini files, 28, 187, 190, 288

database definitions, 256

.iqd files

adding or removing drill-through targets, 126

advantages, 256

as data sources, 199

definition, 365

drill-through targets, 125

setting up drill through, 124

to access server data, 50

using signons, 294

.mdc files, 152

replicating, 138

.mdl files

definition, 365

Using Client-Server Transformer, 177

.mnu files, 23, 28

.ppr file

definition, 365

.psd files, 47

.ps files, 47

.py? files

definition, 365

.qy? files

definition, 365

-(dash)

rsserver command line option, 253

-a command line option, 244

-c

command line option, 245

-cs


-D command line option, 246

-e command line option, 246

-F command line option, 247

-i command line option, 247

-k command line option, 248

-m command line option, 249

-n command line option, 250

-nologo command line option, 250

-o command line option, 250

-ox command line option, 250

-p command line option, 250

-r command line option, 251

-s command line option, 251

-t command line option, 252

-u command line option, 252

-v


-x


AAccess

query source, 48

table source, 48

accessing

cubes, 292

incrementally updated cubes, 286

Access Manager

setting up user class views, 152

signons, 294

user classes, 295

activate

definition, 365

ActiveX Data Object (ADO)

troubleshooting wildcard syntax errors, 285

adding

calculated columns, 61

categories, 22

descendants to categories, 23


dimensions, 20

drill-through files, 124, 127

drill-through targets, 126

extra weeks, 83

extra weeks to synchronize lunar calendars, 207,

213

levels, 20, 62

manual levels, 62, 63

new data sources, 31

partitioning, 286

regular measures, 101

source levels, 63

Additional Reference, 195

adopt

incrementally updated cube data, 163

advanced subset definitions

definition, 365

after rollup

calculated measures, 105

alias

definition, 365

allocated measures

unexpected values after updating time-based parti-

tioned cubes, 297

allocating

as a constant, 207

changing category, 119

changing default, 117

disabling, 118

effect in PowerPlay clients, 292

enabling Allocation tab, 292

measures, 195, 286, 296

measures from levels, 207

measures in time dimensions, 118

proportionally by another measure, 207

setting level, 118

suppressing, 120

allocations

definition, 365

measures, 286

alternate drill-down path

definition, 366

alternate drill-down paths

creating, 65, 195

in time dimensions, 231

level uniqueness, 287

mapping data, 59

partitioning, 287

using dimension views, 232

using with partitioning, 170

analyzing

data flow, 258

source data, 256

ancestor category

definition, 366

ANSI characters, 201

apex

definition, 366

apexing

categories, 150

in alternate drill-down paths, 232

appending

data to cubes, 256

applying

dimension views to cubes, 145

Architect

adding a package to an existing Transformer

model, 180

attributes, 179

filters, 179

packages, 179

using with Transformer, 179

Architect package, 179

creating a Transformer model, 179

using in Transformer, 179

arrays

time, 55

ASCII formats

delimited-field text, 49

fixed-field text, 49

association

definition, 366

attribute

definition, 366

attributes

Architect, 179

authentication

enabling user security in Transformer, 153

updating information in a model, 154

authentication files

updating, 295


Index

AutoDesign

adding a data source, 60

adding levels, 62

creating preliminary models, 60

definition, 366

automatic checks, 140

automatic exceptions

definition, 366

auto-partitioning

modifying, 170

need for hierarchical models, 170

optimizing, 215

autoregression

definition, 367

Bbar chart

definition, 367

base currency

setting up conversion table, 34

base value

definition, 367

batch processing

to create cubes in Transformer (Windows), 250

updating cube connections with cubeswap.bat, 157

before rollup


best practices

model and cube maintenance, 157

model and system design recommendations, 255

strategies to improve build speed, 276

blanks

definition, 367

budgets

degree of detail for dates, 54

buttons

customizing, 25, 26

Ccache memory

adding space to improve cube processing effi-

ciency, 298

calculated categories

creating, 68

defining sets, 70

calculated category

definition, 367

calculated column

definition, 367

calculated columns

adding, 61

defining, 51

calculated measure

definition, 367

calculated measures

after rollup, 105

before rollup, 105

multiplying by 1 to avoid overflow errors, 293

calculation precedence

definition, 367

calculations

adding measures that use if-then-else, 105

creating with the Expression Editor, 239

relative time, 292

Candidate hierarchies pane, 181

categories

adding, 22

adding to partitioned levels, 170

apexing, 220

blocked parent values, 155

business-oriented time, 94

category count, 103

changing allocation, 119

changing the inclusion setting, 205

cloaking, 147, 220

connecting descendants, 23

creating manually, 67

default relative time, 87

deleting if inactive, 161

diagram, 220

disabling refresh option, 168

displaying number in dimensions and levels, 22

excluding, 205, 220

generating, 97, 263, 288

generating date, 85

generating using a specific data source, 98

including, 205

invalid dates, 289

manually-created orphanages, 99

moving, 22, 160

omitting, 146, 147, 148


Index

omitting ancestors and siblings, 150

omitting descendants, 150

ordering, 71

ordering for the entire model, 72

orphans of, 97, 288

preventing new, 100

relative time, 74, 87, 292

server, 198

setting period-to-date, 208

simultaneously disabling creation of categories and

cube rebuild, 250

sorting in time-based partitioned cubes, 286

summarizing, 220

suppressing, 170, 205, 220

troubleshooting slow generation, 288

troubleshooting unbalanced hierarchies, 288

category

definition, 368

category codes

definition, 368

category counts

defining as a measure, 103

definition, 368

category label

definition, 368

category labels

incorrect decimal placement (Oracle), 294

Category property sheet

share category, 186

category sets

definition, 368

category viewer

adding levels, 20

definition, 368

deleting levels, 20

overview, 22

century

handling 21st century dates, 57, 97

centurybreak

Trnsfrmr.ini option, 57

centurybreak Trnsfrmr.ini setting, 190

cerlocale.xml, 300

cern.ini file, 187

locale type setting, 190

change function, 239

changing

data sources, 292

log file parameters, 191

characters

field delimiters, 201

forward slash for UNIX directory paths, 203

sets, 201

wildcard, 219

character styles, 388

checking

cube status, 143, 263

model for potential problems, 100, 262

Check Model

troubleshooting command, 262

checkpoints

for failed models, 158

child category

definition, 368

children

definition, 369

clean house

deleting inactive categories, 161

client-server

overview, 177

server processes, 198

synchronizing models, 197

troubleshooting connection problems, 299

Clipper, 48, 199

cloak

definition, 369

cloaking

categories, 147


clustered bar chart

definition, 369

Cognos.ini file, 187, 288

cognos security administration file

definition, 369

collection

definition, 369

colors

scope in dimension map, 195

column

definition, 369

columns

adding calculated, 61


Index

associating with levels, 234

calculated, 51

defining, 52


matching source and model, 159

modifying, 292

naming, 51

overview, 51

synchronizing, 292

time arrays, 55

combining

dimension and user class views, 156

command line options, 241

- (dash), 253

-a, 244

-c, 245

-cs, 245

-D, 246

-e, 246

-F, 247

-i, 247

-k, 248

-m, 249

-n, 250

-nologo, 250

-o, 250

-ox, 250

-p, 250

-r, 251

-s, 251

-t, 252

-u, 252

-v, 253

-x, 253

Common measures pane, 181

comparing

performance in different time periods, 94

Transformer editions, 15

compatibility

using MDL to move between versions, 31

compressing

cubes, 141

condition

definition, 369

configuring

data formats, 300

Connect.ini file, 288

connecting

client-server computers using PowerGrid, 299

descendant categories, 23

consolidating

data in cubes, 167

overview, 120

reducing data records, 257

summary partitions, 170

with duplicates rollup, 215

consolidation

definition, 369

purpose and limitations, 124

constant

definition, 370

constraints

unsupported database joins, 258

control cube

definition, 370

controlling

adding extra weeks for lunar, 83

range of dates, 84

rendition registry, 191

splitting weeks, 81

when measures are calculated, 105

controlling data access

based on dimension and user class views, 156

to PowerCubes in IBM Cognos 8, 283

conventions

quotation marks, 196

convergence levels

definition, 370


conversion

currency overview, 33

euro triangulation method, 36

non-EMU currencies, 34

setting up base table, 34

setting up euro table, 36

converting

currencies for specified measures, 39

copying

.mdl files to the server, 177

correlation display

definition, 370


Index

counts

categories per level, 262

Create Selected Cubes

command, 263

creating

.mdc files, 138

alternate drill-down paths, 65

calculated categories, 68

categories manually, 67

cubes, 141

cubes using multiple files, 170

cube with externally rolled up measure values, 109

currency tables manually, 39

dimensions, 60

dimension views, 145

levels with special categories, 96

models that use multiple data sources, 167

multifile cubes, 174

new models, 30

preliminary model designs, 60

prototype server models, 177

selected cubes only, 263

server categories and cubes, 198

special categories, 95

special category levels, 96

test cubes, 142

time dimension manually, 77

time dimensions, 76

Transformer model from an Architect package, 179

unbalanced hierarchies within a dimension, 73

crosstab chart

definition, 370

cube

definition, 370

cube connections

updating using cubeswap.bat, 157

cube creation

data sources, 140

cube group

definition, 370

cube object

definition, 370

cubes

accessing from PowerPlay clients, 292

access time, 286

adopting incrementally updated data, 163

algorithm for building, 170

allowing for future group members, 161

appending data, 256

automatic checks, 140

building algorithms, 170

check status, 143, 198, 263

creating, 141

creating multifile cubes, 190

creating selected cubes only, 263

creating test cubes, 142

creating using multiple files, 170, 174

custom views, 144

defining groups, 130

defining manually, 129

disabling creation, 250

enabled/disabled, 198

excluding measures, 138

improving access speed, 162

leaving data unchanged, 109

maintaining, 157

manually customizing to meet specific user

needs, 129

multiple storage locations, 256

omitting dimensions, 138

opening at non-root level, 74

optimal partitioning strategies, 168

optimizing creation, 256

optimizing processing, 165, 215

protecting, 152, 283

replicating, 138

server, 198

setting passwords in Series 7, 139

sorting constraints for time-based partitioned

cubes, 286

splitting, 170

time-based partitioning, 130

time-based partitioning customization, 136

troubleshooting slow build times, 276

UNIX, 127

updating incrementally, 163, 256, 286, 288

updating metadata only, 142

currency

conversion overview, 33

default conversion table, 34

euro conversion table, 36

setting up manual conversion, 39


Index

currency records

definition, 371

currency table

definition, 371

currency tables

creating manually, 39

current periods

definition, 371

details, 88

setting automatically, 88

setting manually, 88

setting on command line, 252

setting the current period, 208

custom exceptions

definition, 371

customizing

buttons, 25

cube content using views, 144

cubes to meet specific user needs, 129

launch buttons, 25, 27

launch menu commands, 25

menus, 23

time dimensions, 74

toolbar buttons, 25, 26

toolbars, 25, 26

custom time dimensions

definition, 371

Ddata

alternative source files, 256

appending to cubes, 256

consolidating, 257

ensuring integrity, 258

flow analysis, 258

how volume affects performance, 258

retaining historical data, 256

separating transactional and structural sources, 257

static and non-static sources, 256

structuring, 258

summarizing, 257

data access

controlling with dimension and user class views, 156

database joins

troubleshooting prior to data import, 258

databases

accessing from Impromptu, 256

building pointer files, 138

isolation levels, 256

reboot required after configuring a new connec-

tion, 298

references in .ini file, 256

supplying password, 248, 249

data dictionary

using to resolve differences, 258

data formats, 300

configuring, 300

data item

definition, 371

data source

definition, 371

data source directories

definition, 371

data sources

.iqd files, 50

adding new, 31

ASCII text formats, 49

associating with alternate files, 140

changes in structure, 160

changing, 292

currency conversion, 34

customizing, 140

defining columns, 51

defining multiple sources for cubes, 296

defining rates for currency conversion, 33

defining rates for euro currency conversion, 36

fixed-field text, 52

inclusion of categories in cubes, 205

local, 44

local databases, 48

maintaining, 157

matching to model columns, 159

multiple, 167, 296

PowerHouse Portable Subfiles, 47


relation to dimension map, 234

scope, 195

server, 48, 50

specifying alternates when creating cubes, 140

specifying before generating categories, 98

specifying different source files, 256


Index

spreadsheet, 44

structural, 257

supported types, 44, 199

updating metadata, 142

using Modify Columns command to synchronize

changes, 292

viewing scope, 40

data source scope

show on a dimension map, 21

data source scope maps

definition, 372

Data Source Viewer

using to modify SQL in .iqd files, 32

data types

changing to avoid decimal and precision errors, 293

problems importing BLOB and CLOB query

items, 298

date_exp

definition, 372

date columns

degree of detail, 212

date ranges

limitations, 84

dates


creating dimensions manually, 77

defining degree of detail, 54

defining early dates, 97

defining input format, 76

defining late dates, 97

defining non-standard dates, 85

format codes, 85, 204

functions, 213

generating categories, 85

handling invalid, 98

in alternate drill-downs paths, 231

input formats, 203

limiting range of dates, 84

setting up calendar years, 79

setting up fiscal, 79

setting up fiscal dates, 79

setting up lunar dates, 82

setting up time dimensions, 74

splitting weeks, 81

troubleshooting invalid dates, 289

using Date wizard, 76

using time arrays, 55

Year 2000, 57, 97

date styles, 388

datetime_exp

definition, 372

Date wizard

using to create time dimensions, 76

dBase, 48, 199

decimal places

measures, 293

scaling, 53

troubleshooting errors, 293

defining

calculated columns, 51


cube groups, 130

cubes manually, 129

date format codes, 204

date formats, 203

date input, 76

if-then-else calculated measures, 105

partitions manually, 172

scope, 195

security requirements, 255

server models, 177

sets for calculated categories, 70

time-based partitioned cubes, 130

degree of detail

date columns, 212

deleting

categories, 22

dimensions, 20

inactive categories, 161

levels, 20

delimited field

data sources, 199

delimited fields

delimiting characters, 201

delimited field text files

ASCII text formats, 49

definition, 372

delimiters

field, 201

quotation marks in MDL, 196

delimiting

fields, 201


Index

denied values

blocked parent totals if children excluded, 155

descendant category

definition, 372

designing

Transformer using the Metadata Explorer, 182

diagrams

category actions, 220

definition, 372

overview, 22

zooming, 22

dimension

definition, 372

dimension diagrams

definition, 372

overview, 22

dimension line

definition, 372

dimension map

adding dimensions, 20

deleting dimensions, 20

how data sources relate, 234

overview, 19

sample pilot project, 259

scope, 21, 195

show category counts, 22

dimension maps

definition, 373

dimension menus

definition, 373

dimensions

adding, 20

adding source levels, 63

automatically creating for dates, 76

creating new, 60

creating one time dimension per cube, 60

degree of detail for time dimension, 54

deleting, 20

locking out new categories, 100

mapping data into multiple dimensions, 59

omitting from cubes, 138

omitting when using auto-partitioning, 174

opening at non-root level, 74

overview, 59

setting allocation, 117

setting up date and time dimensions, 74

showing, 59

splitting weeks, 81

with alternate drill-down paths in time dimen-

sions, 231

dimension table

definition, 373

dimension viewer

definition, 373

dimension views

alternate drill-down paths, 232

applying to cubes, 145

combining with user class views, 156

creating, 145

definition, 373

using to customize cubes, 144

directory paths

forward slash for UNIX delimiters, 203

directory server

definition, 373

disabling

refresh option, 168

rollup on special categories, 96

disasters, 191

display

definition, 373

displaying

missing values, 107

number of categories, 22

distributing

cubes, 138

custom menus, 28

custom toolbars, 28

DOS Code Page 437, 201

drill

definition, 373

drill categories

definition, 374

drill down

definition, 374

drill down paths

definition, 374

drill-down paths

creating, 65, 195

in time dimensions, 231

mapping data, 59

special (alternate) categories, 96


Index

using dimension views, 232

drill through

cube-to-cube, 125

definition, 374

Impromptu, 125

third-party OLAP source, 125

to a measure, 124

to any associated file, 124, 126, 127

to Essbase Linked Partitions, 124, 126

to Essbase Reporting Objects, 124, 126

to IBM Cognos 8 using IBM Cognos Connec-

tion, 124

to Impromptu, 124

to Impromptu reports, 124, 126, 127

to other PowerPlay reports and cubes, 124, 126

to other UNIX cubes, 127

to ReportNet reports, 124

using .iqd files, 125

drill through targets

uniqueness issues, 236

drill-through targets

adding, 126

removing or excluding by level, 126

drill up

definition, 374

driving categories

definition, 374

duplicate records

consolidation criteria, 257

definition, 374

duplicates

rollup, 122

Eearly dates

handling, 57, 97

editions, 15

eliminating

source or key orphanages, 97

embed

definition, 374

empty result sets

troubleshooting wildcard syntax errors, 285

enabling

currency conversion, 39

measure allocation, 292

user security for a model, 153

entities

filtering, 183

filtering in the business layer, 183

entity attributes

filtering, 183

environment variables, 203

changing default timeouts, 299

errors

help for specific error messages, 303

help in resolving, 263

troubleshooting documentation, 263

euro

setting up currency conversion, 33, 36

Excel, 44, 199

exception dimensions

adding, 61

exception highlighting

definition, 375

exclude

definition, 375

excluding

categories, 148, 205

child categories, 155

drill-through targets by level, 126


measures, 138

explain window

definition, 375

Explorer reports

definition, 375

exported model files

definition, 375

expression

definition, 375

Expression Editor

help on available functions, 239

expressions

creating with Expression Editor, 239

unique to Transformer, 239

externally rolled up measures

creating, 109

extra weeks, 207


Index

Ffact tables

definition, 375

failed processes

model recovery, 158

fatal errors

failed model recovery, 158

feelers

definition, 375

file types

.asc files, 49

.crr files, 124

.csv files, 49, 201

.imr files, 124, 126, 127

.ini files, 190, 288

.iqd files, 50, 124, 126, 127, 199

.log files, 191

.mdb files, 199

.mdc files (Series 7 PowerCubes), 152

.mdl files, 31, 177

.mdp files, 190

.psd files, 47

.ps files, 47

.vcd files, 137

filter

definition, 376

filter bars

definition, 376

filtering

Architect data source in Transformer, 183

entities for a user class, 183

entities in the business layer, 183

entity attributes for a user class, 183

subjects, 183

filters

Architect, 179

find-in-cube subset definitions

definition, 376

Find in Report

definition, 376

fiscal dates

setting up, 79

fixed field

data sources, 199

fixed field text files

ASCII format, 49

fixed-field text files

defining columns, 52

definition, 376

forecast

definition, 376

formats

data, 300

date codes, 204

date input, 203

defining date input, 76

defining for date values, 85

FoxPro, 48, 199

function

definition, 376

functions

change, 239

creating with Expression Editor, 239

percent-growth, 239

share, 240

unique to Transformer, 239

GGenerate Categories

command, 263

generating

categories, 97, 263

categories using a specific data source, 98

date categories, 85

GPF

recovering from General Protection Faults, 158

See Also troubleshooting

growth

definition, 376

Hhandling

early and late dates, 97

invalid dates, 98

hiding

root-level values in scenario dimensions, 74

toolbars, 25

hierarchies

unbalanced, 288

horizontal axis

definition, 377


Index

IIBM CognosScript language

definition, 377

IBM Cognos security administration files (.csa)

definition, 377

identify the measures, 101

importing

.mdl model files from earlier versions, 31

Impromptu

accessing databases, 256

advantages of using .iqd files, 256

drill-through, 124

drill through from PowerPlay, 124

drill-through reports, 126, 127

drill through without .iqd files, 125

using to prepare Transformer data sources, 256

using to query local and server data, 50

Impromptu Query Definition

definition, 377

inactive categories

deleting, 161

including

categories in cubes, 205

inclusions, 205

incremental updates

definition, 377

limitations, 164

optimize using, 162, 163, 256, 286, 288

indented layouts

definition, 377

information messages, 191

initialization files

Cognos.ini, 288

Connect.ini, 288

Trnsfrmr.ini and cern.ini, 190

input scales

avoiding truncated data values for migrated meas-

ures, 293

incorrect decimal placement in numeric category

labels, 294

setting, 53

truncated decimal values for measures, 293

integer_exp

definition, 377

intersected categories

definition, 377

interval_exp

definition, 377

invalid dates

handling, 98

isolation levels

overriding for databases, 256

setting for Impromptu Query Definition files

(.iqd), 202

Jjoining tables

troubleshooting, 288

joins

troubleshooting prior to data import, 258

Kkey orphanages

eliminating, 97

Llate dates

handling, 57, 97

launch buttons

setting up, 27

launch menu commands

setting up, 25

layer

definition, 378

leaf category

definition, 378

legend

definition, 378

level

definition, 378

level headings

definition, 378

levels

adding source, 63

adding to dimensions, 20, 62

allocating in time dimensions, 118

changing the category inclusion setting, 205

deleting from dimensions, 20

derived, 195

isolation, 202

manual, 62, 63

ordering categories within, 71


Index

setting allocation, 118

unique, 212, 287, 289

uniqueness, 234

with special categories, 96

Level tab, 181

limitations

in consolidation, 120, 124

in range of dates, 84

when partitioning, 168

line chart

definition, 378

link

definition, 378

lists

definition, 378

for Transformer objects, 19

local authentication

definition, 378

local authentication export file (.lae)

definition, 379

local currency styles, 388

local data

overview, 44

local databases

data sources, 48

locale

settings, 300

locale settings

configuring, 300

localetype cern.ini setting

date and numeric display formats, 190

locked dimensions

definition, 379

locked levels

definition, 379

locking

dimensions, 100

log file

monitoring partitioning impacts, 170

LOG File

definition, 379

log files

manual partitioning based on records processed per

pass, 264

logging, 191, 251, 252

long names

definition, 379

Lotus 1-2-3, 44, 199

lowest details

definition, 379

lunar time

adding extra weeks, 83

adding extra weeks to resynchronize, 207

setting up, 82

Mmacro

definition, 379

macros

using a menu command, 25

using a toolbar button, 27

maintaining

cubes, 157


data sources, 157

models, 157

source data, 157

updated cube connections, 157

manual categories

adding descendants, 23

creating, 67

manual levels

adding, 63

definition, 380

manual partitioning

example, 264

reasons for using, 172

mapping

data into dimensions, 59

markers

definition, 380

matching

model and source columns, 159

MDL (Model Definition Language)

definition, 380

MDL syntax

quotation mark delimiters, 196

measure

definition, 380

measure allocation

in models with multiple data sources, 40


Index

measure folder, 102

definition, 380

measures

adding, 103

adding calculated, 104

adding if-then-else calculated, 105

adding regular, 101

allocating, 115, 296

avoiding decimal and precision errors, 293

avoiding overflow errors, 293

decimal values, 293

drill through, 124

excluding from cubes, 138

externally rolled up, 109

grouping, 138

ignoring specified null or missing time state rollup

values, 337

measure folders, 102

missing values in PowerPlay, 291

NA or zero values, 291

not rolled up, 109

overview, 108

regular, 113

regular rollup, 108

reversing the sign of values, 107

rollup, 103, 108

scaling values, 53

scope, 195

time state, 113

time state rollup, 111

timing, 105

unsummarized, 109

zero values, 291

measure scope map

definition, 380

showing scope on user interface, 21

Measure tab, 181

measuring

past performance, 89, 91

past performance to-date, 90

members, See categories

member unique names, See category codes

memory

adding space to improve cube processing, 298

memory allocation

troubleshooting startup problems by setting

OS_AS_SIZE, 297

menus

adding commands, 23

creating, 23

customizing, 23

distributing, 28

resetting, 23

merging

data from different sources, 258

messages

help for specific errors, 303

metadata

definition, 380

updating, 142

metadata explorer

Transformer, 179

Metadata Explorer, 181

Candidate hierarchies pane, 181

Common measures pane, 181

designing a Transformer model, 182

level tab, 181

measure tab, 181

package view, 181

Show Model, 181

Show Scope, 181

Subject tab, 181

Microsoft Access

query source, 48

table source, 48

mirror reference

definition, 380

missing values

definition, 380

displaying, 107

ignoring for specified time state rollups, 113

measures, 291

model

definition, 381

modeling

design recommendations, 255

model prototype

example and dimension map, 259


Index

models

adding an Architect package to an existing

model, 180

adding an Architect package to Transformer, 180

adding data sources, 31

avoiding problems with non-unique category

codes, 165

changes in structure, 159

checking, 198

checking for potential problems, 100, 262

converting, 31

creating from an Architect package, 179

creating new, 30

creating server prototypes, 177

creating with AutoDesign, 60

designing, 182

enabling user class views, 153

importing, 31

maintaining, 157

multiple data source type, 258

preparing for partitioning, 170

recovering after processing failure, 158

setting up multiple sources, 40

synchronizing, 177

synchronizing client-server, 197

uploading files to the server, 177

using different source files, 256

using multiple data sources, 256

modifying

auto-partitioning, 170

columns, 292

SQL in .iqd files (Data Source Viewer), 32

monthly time arrays

steps and examples in different model types, 55

moving

categories, 22, 160

MS Access, 199

multi-byte characters, 201

multifile PowerCubes

creating, 170

multiline chart

definition, 381

multiple data sources

advantages, 256, 258

data source scope, 195

with time arrays, 55

multiple file cubes

threshold size setting, 190

multiple source models

overview, 40

setting up, 40

Nnamespace

definition, 381

namespaces

configuring security in models, 153

NA or zero values

missing measures, 291

nested category

definition, 381

nested charts

definition, 381

nested crosstab

definition, 381

nested measures, 102

NetInfo

definition, 381

utility used with Transformer for UNIX, 177

New Data Source wizard

using to add data sources to existing models, 31

New Model wizard

using to define new models, 30

NewsBox

definition, 381

NewsItem

definition, 381

null values

displaying, 107

ignoring for specified time state rollups, 113

number styles, 388

numeric_exp

definition, 381

Oobject identifiers

definition, 382

Object Linking and Embedding

definition, 382

ODBC, 48

OLE

definition, 382


Index

omitting

ancestor and sibling categories, 150

categories using Cloak, 147

categories using Exclude, 148

categories using Summarize, 150

categories using Suppress, 146

descendant categories, 150

dimensions from cubes, 138

dimensions using auto-partitioning, 174

opening

dimension diagrams, 22

operator

definition, 382

optimization

case studies, 276

optimization tests

sample model description, 268

optimizing

cube processing, 165, 215

disabling Refresh, 168

PowerCubes to open faster in IBM Cognos 8, 283

query performance, 162

using consolidation, 167

using multiple sources, 167

ordering

categories using global preference setting, 72

categories within levels, 71

orphanages

definition, 382

setting up manual levels, 99

orphans

categories, 97, 288

outer joins

using for data integrity, 258

output scales

setting, 53

updating MDL model to match datasource columns

and measures, 253

overflow values

definition, 382

Ppackage, Architect, 179

package, using in Transformer, 179

packages

Architect, 179

package view, 181

Paradox, 48, 199

parentage subset definitions

definition, 382

parent categories

blocked total values, 155

parent category

definition, 382

partial weeks, 81, 207

partitioning

adding new categories, 170

adding new levels, 286

alternate drill-downs, 287

automatic method, 170

based on information in log file, 264

category sort after a time-based partitioned cube

merge, 286

definition, 383

log file monitoring, 170, 252

manual method, 170, 172

passes, 170, 172

performance tradeoffs, 162

preparing models, 170

process for devising an optimal strategy, 168

resetting, 172

retaining sort order, 286

strategies and limitations, 168

strategy, 172

troubleshooting poor query performance, 300

using incremental updates, 286

using with alternate drill-down paths, 170

partitions

consolidation, 170

definition, 383

size, 170

summary, 170

passes

partitioning, 170, 172

passwords

securing Series 7 PowerCubes, 139

paths, 203

pattern matching, 219

PDF File

definition, 383

percent-growth function, 239


Index

performance

analyzing for periods-to-date, 90

analyzing for relative time periods, 89

analyzing over time, 91

improving by disabling uniqueness verification, 165

improving by normalizing data before import, 258

improving PowerCube open times in IBM Cognos

8, 283

performance indicator

definition, 383

period-to-date categories, 208

Personal Transformer, 15

pie chart

definition, 383

pilot project

example and dimension map, 259

placeholder categories

adding, 161

planning

for new cube group members, 161

plans


Pointer File

definition, 383

pop-up menus

available in Transformer, 19

Portable Document Format File (.pdf)

definition, 383

portable subfile dictionary files, 47

PowerCube

definition, 383

PowerCube groups

definition, 383

PowerCube objects

definition, 383

PowerCubes

adopting incrementally updated data, 163

checking status, 143

check status, 198

creating, 141

creating multifile cubes, 190

creating test cubes, 142

creating using multiple files, 174

custom views, 144

defining groups, 130

defining manually, 129

disabling creation, 250

enabled/disabled, 198

excluding measures, 138

maintaining, 157

measure folders, 138

omitting dimensions, 138

optimizing processing, 165, 215

planning new group member, 161

protected, 152, 283

replicating, 138

server, 198

setting passwords, 139

speeding up access, 162

temporary filenames, 140

troubleshooting slow build times, 276

updating connections using cubeswap.bat, 157

updating incrementally, 163

updating metadata, 142

PowerCube Status

command, 263

PowerGrid

Connect.ini file, 288

definition, 383

network daemon used with Transformer for

UNIX, 177

using for client-server mode, 288

PowerHouse Portable Subfiles

data sources, 47

PowerPlay

database connect utility, 138

drill through to other cubes, 126

drill through to other reports and cubes, 124

drill through to UNIX cubes, 127

editions, 15

uniqueness issues when drilling cube-to-cube, 236

PowerPlay Portable Report File (.ppx)

definition, 383

PowerPlay Web Explorer

definition, 384

PowerPlay Web Viewer

definition, 384

PowerPlay Worksheets

definition, 384

Ppconnct.exe

connecting to remote cubes, 138


Index

precedence

rolling up calculated measures, 105

precision

troubleshooting errors, 293

troubleshooting overflow errors, 293

preferences

specifying files, 247

Trnsfrmr.ini and cern.ini, 190

preprocessing

normalizing data to improve performance, 258

source data, 257

presummarization

definition, 384

primary drill-down paths

definition, 384

processing

adding cache space to improve efficiency, 298

optimizing, 165

production reports


property sheet

Category, 186

Level, 186

share category, 186

share object, 186

Special Category, 186

property sheets

for Transformer objects, 19

Protected Cube

definition, 384

protected cubes

overview, 152

prototypes

creating preliminary designs, 60

transferring from client to server, 177

proven practices

adding faster application servers, 269

allocating more temporary file space, 271

enabling multiprocessing, 269

limiting model and cube size, 276

model and system design, 255

optimizing environment settings (UNIX), 273

optimizing environment settings (Windows), 272

optimizing gateway settings, 274

optimizing memory allocation, 270

redistributing data files, 273

reorganizing data folders, 271

resolving processing bottlenecks, 275

strategies to improve build speed, 276

testing optimization strategies, 268

Qquarterly time arrays

steps and examples in different model types, 55

query

definition, 384

Query Processor

definition, 384

Quick Tour

introduction to PowerPlay, 15


Rrank

definition, 384

rank categories

definition, 384

recovering

failed models, 158

references

source for each column in model, 262

registering

base key information, 191

regression

definition, 385

regular rollup

setting, 108

relative time

calculations, 292

categories, 87, 88

future projections, 92

settings, 208

relative time category

definition, 385

relative time functions

definition, 385

removing

drill-through files, 124

unsupported database joins, 258

renditions

.ini files, 191

registering base keys, 191


Index

Trnsfrmr.ini location, 190

reporter reports

definition, 385

Report Processor

definition, 385

reports

definition, 385

production, 300

resetting

menus, 23

partitions, 172

toolbars, 25

resolving

data dictionary terms, 258

errors using online Help, 263

restricting

data access based on user class, 154

drill-through targets by level, 126

retaining

combined rollup values, 150

rollup values (Cloak), 147

reversing

signs, 107

roles

definition, 386

rollup

category count, 103

combining regular and time state, 113

controlling measures, 108

disabling on special categories, 96

duplicates, 120, 122

external, 109

ignoring null or missing values, 337

in calculated measures, 105

in consolidation, 120

measure folder, 102

overview, 108

regular, 108, 113, 120

retaining values for summarized categories, 150

time state, 111, 113

rollup calculations

definition, 386

rollup functions

definition, 386

rollups

duplicate records, 257

ignoring null and missing values, 113

rollup types

supported functions, 215

root category

definition, 386

rounding

scaling, 53

row

definition, 386

Sscales

updating MDL model to match datasource columns

and measures, 253

scaling

input values, 53

measure values, 53

scaling errors

in numeric category labels, 294

scatter plot chart

definition, 386

scenario dimensions

hiding the root category, 74

scope

defining, 195

show on dimension map, 21

show option for measures and categories, 262

Scope

function, comparing in Transformer and Archi-

tect, 184

searching

using wildcards, 219

security

augmenting for PowerCubes in IBM Cognos 8, 283

combining dimension and user class views, 156

updating, 295

updating access-controlled views, 154

using for custom views in Transformer, 153

security requirements

defining, 255

server

accessing data, 50

category and cube processes, 198

copying .mdl files, 177, 190

creating prototypes, 177

data sources, 50


Index

defining models, 177

model synchronization, 197

synchronization with client, 197

uploading models, 177

servers

connection problems, 299

setting

allocation type for a measure, 207

category inclusion, 205

cube processing optimization, 215

date functions, 213

degree of detail for date columns, 212

partitions manually, 172

relative time periods, 208

settings

locale, 300

setting up

calendar and fiscal years in one time dimension, 79

calendar years, 79

currency conversion, 33, 34, 39

currency conversion tables manually, 39

current period automatically, 88

current period manually, 88

custom time dimensions, 85

drill through, 124

euro currency conversion, 36

fiscal years, 79

future time projections, 92

launch buttons, 27

launch menu commands, 25

lunar time periods, 82

months, 79

orphanages as manual levels, 99

protected cubes, 152

quarters, 79

relative time categories, 87

special time categories, 94

time dimensions, 74

time state rollup function, 111

unique levels in drill-through targets, 236

share

category, 186

object, 186

property sheet, 186

shared dimensions

definition, 386

share function, 240

short names

definition, 386

Show Counts

command, 262

showing

toolbars, 25

Show Model, 181

Show References

command, 262

Show Scope, 181, 184

command, 262

sibling category

definition, 387

signons

in models, 294

using with .iqd files, 294

signs

reversing, 107

size

partitions, 170

slice and dice

definition, 387

slowly changing dimensions, 134, 135

sort

definition, 387

sorting

categories in time-based partitioned cubes, 286

sort order

retaining in time-based partitioned cubes, 286

source data

analyzing, 256

preprocessing, 257

requirements, 256

source files

alternative, 256

source orphanages

definition, 387

eliminating, 97

source values

definition, 387

sparse data

definition, 387

special categories

adding descendants, 23

creating, 95


Index

creating levels, 96

definition, 387

disabling rollup, 96

overview, 94

Special category property sheet

share category, 186

specifying

partitioning passes, 170, 172

partition size, 170

uniqueness, 289

splitting

cubes into multiple files, 170

splitting weeks, 81, 207

spreadsheet crosstabs

definition, 388

spreadsheet databases

definition, 388

spreadsheets, 44

SQL

modifying for .iqd files (Data Source Viewer), 32

stacked bar chart

definition, 388

standard layouts

definition, 388

standard reports

definition, 388

status

check model, 198

of cubes, 143, 198

storage types

changing to avoid overflow errors, 293

definition, 388

stored procedure

definition, 388

strategies

cube partitioning, 168

string

definition, 388

string_exp

definition, 388

structural data sources

definition, 388

separating from transactional, 257

static data, 257

Structured Query Language, See SQL

style

definition, 388

subdimensions

creating, 73

definition, 389

subject

definition, 389

subject item

definition, 389

subjects

filtering, 183

Subject tab, 181

subset calculations

definition, 389

subset definitions

definition, 389

subsets

definition, 389

subset viewer

definition, 389

summaries

values in PowerPlay clients, 292

summarize

definition, 390

summarizing

categories, 150


reducing data processing, 257

summary

definition, 390

summary partitions

definition, 390

suppress

definition, 390

suppressing

allocation, 120

allocation for a measure, 207

categories, 146, 205


unwanted manually added categories, 170

suspended models

recovering, 158

swapping

definition, 390

switches, See command line options


Index

synchronization

definition, 390

of client and server models, 177

synchronizing

client-server models, 197

column and measure scales in MDL model with

datasource, 253

columns with new data, 292

syntax

quotation marks in MDL, 196

Ttable aliases

using for data integrity, 258

tables

joining, 288

test cubes

creating, 142

third-party OLAP source

drill through, 125

ticks

definition, 390

time


allocating measures, 118

arrays, 55

comparing past and present performance, 89, 90,

91

create dimension manually, 77

current period, 233

custom, 74

default relative categories, 87

dimensions, 74, 77, 79, 85

invalid dates, 289

lunar periods, 82

measuring data in relative, 89, 90, 91

non-standard, 85

period-to-period comparisons, 94

projections, 92

setting current period automatically, 88

setting current period manually, 88

setting up lunar, 82

setting up projections, 92

setting up relative, 87

splitting partial weeks, 207

time_exp

definition, 390

time arrays

definition, 391

monthly, 55

quarterly, 55

time-based partition cube

definition, 391

time-based partitioned cube

adding or removing cubes, 137

multi-level cube, 137

slowly changing dimensions, 134, 135

time-based partitioning

cube customization, 136

defining, 130

limitations of this cube design, 130

retaining sort order, 286

time dimensions

definition, 391

invalid categories, 289

time interval styles, 388

time levels

definition, 391

timeout settings

changing environment variable defaults, 299

time state measures

definition, 391

time-state measures

ignoring specified null or missing rollup values, 337

time-state rollups

ignoring null and missing values, 113

time styles, 388

timing

setting for calculated measure, 105

toolbar buttons

customizing, 25

toolbars

changing buttons, 26

customizing, 25, 26

distributing, 28

resetting, 25

setting up buttons, 26

showing or hiding, 25

TR0104, 303

TR0105, 303

TR0106, 303


Index

TR0107, 304

TR0108, 304

TR0109, 305

TR0110, 305

TR0111, 305

TR0112, 305

TR0113, 306

TR0114, 306

TR0116, 306

TR0117, 306

TR0118, 307

TR0128, 308

TR0131, 308

TR0132, 308

TR0133, 309

TR0137, 309

TR0149, 310

TR0151, 310

TR0202, 310

TR0203, 310

TR0205, 311

TR0206, 311

TR0207, 311

TR0208, 311

TR0209, 311

TR0210, 312

TR0214, 312

TR0215, 312

TR0217, 313

TR0301, 313

TR0303, 313

TR0404, 313

TR0407, 313

TR0408, 314

TR0412, 314

TR0420, 314

TR0423, 314

TR0476, 315

TR0500, 315

TR0501, 315

TR0502, 315

TR0503, 316

TR0504, 316

TR0505, 316

TR0507, 317

TR0508, 317

TR0510, 317

TR0514, 317

TR0515, 318

TR0518, 318

TR0519, 318

TR0523, 318

TR0524, 319

TR0525, 319

TR0528, 320

TR0534, 320

TR0535, 320

TR0536, 320

TR0538, 321

TR0540, 321

TR0541, 321

TR0552, 322

TR0605, 322

TR0606, 322

TR0607, 323

TR0613, 323

TR0621, 323

TR0623, 324

TR0624, 324

TR0700, 324

TR0701, 325

TR0702, 325

TR0705, 325

TR0713, 325

TR0716, 326

TR0743, 326

TR0745, 326

TR0749, 327

TR0750, 327

TR0751, 327

TR0752, 327

TR0753, 327

TR0800, 327

TR0802, 328

TR0803, 328

TR0804, 329

TR0805, 329

TR0806, 329

TR0807, 329

TR0808, 329

TR0809, 330

TR0810, 330


Index

TR0811, 330

TR0812, 331

TR0813, 331

TR0815, 331

TR0816, 331

TR0817, 332

TR0900, 332

TR0904, 332

TR0906, 332

TR0907, 333

TR0914, 333

TR0917, 333

TR0919, 333

TR1003, 334

TR1005, 334

TR1100, 334

TR1101, 334

TR1102, 334

TR1106, 335

TR1109, 335

TR1304, 335

TR1307, 335

TR1308, 336

TR1309, 336

TR1310, 336

TR1312, 336

TR1319, 337

TR1320, 337

TR1340, 337

TR1502, 338

TR1503, 289, 338

TR1601, 338

TR1700, 338

TR1703, 339

TR1900, 339

TR1903, 339

TR1907, 340

TR2000, 340

TR20002, 341

TR2001, 340

TR2306, 341

TR2307, 341

TR2308, 341

TR2312, 289, 342

TR2313, 289, 342

TR2314, 289, 342

TR2316, 343

TR2317, 289, 343

TR2318, 289, 343

TR2319, 289, 344

TR2320, 344

TR2321, 345

TR2322, 345

TR2323, 345

TR2324, 345

TR2325, 346

TR2326, 346

TR2502, 346

TR2503, 347

TR2504, 347

TR2505, 347

TR2506, 347

TR2507, 348

TR2508, 348

TR2509, 348

TR2510, 348

TR2600, 349

TR2601, 349

TR2605, 349

TR2606, 349

TR2608, 350

TR2700, 350

TR2701, 350

TR2702, 351

TR2703, 351

TR2704, 352

TR2705, 352

TR2706, 352

TR2707, 352

TR2708, 353

TR2710, 353

TR2711, 353

TR2712, 354

TR2713, 354

TR2714, 354

TR2715, 354

TR2716, 355

TR2717, 355

TR2718, 355

TR2719, 356

TR2720, 356

TR2723, 356


Index

TR2724, 356

TR2725, 357

TR2726, 357

TR2727, 357

TR2728, 358

TR2729, 358

TR2731, 358

TR2732, 358

TR2733, 359

TR2734, 359

TR2735, 359

TR2736, 359

TR2737, 360

TR2740, 360

TR2800, 360

TR2801, 360

TR2802, 361

TR2803, 362

TR2804, 362

TR3006, 362

TR3010, 363

TR3014, 363

TR3101, 363

TR3311, 363

TR3439, 364

TR3505, 364

TR3619, 364

transactional data sources

definition, 391

separating from structural, 257

Transformer

creating a model from an Architect package, 179

designing, 182

Metadata Explorer, 179

using with Architect, 179

Transformer Edition, 15

Transformer model

adding an Architect package to an existing

model, 180

trend

definition, 391

triangulation

definition, 391

euro conversion method, 36

overview, 33

Trnsfrmr.ini file

century break setting, 190

troubleshooting

client-server connection problems, 299

cube connection problems using cubeswap.bat, 157

database connection problems, 298

empty results when querying Microsoft Access

databases, 285

errors arising from use of double quotation

marks, 285

help for specific error messages, 263

import problems with BLOB and CLOB data

types, 298

incorrect scaling of numeric category labels, 294

measure allocation anomalies in updated time-based

partitioned cubes, 297

model and system design issues, 255

numeric data migration errors, 293

overflow errors, 293

problems due to non-unique category codes, 165

production problems, 276

query performance for partitioned cubes, 300

slow PowerCube open times in IBM Cognos 8, 283

startup problems due to ObjectStore VirtualAlloc

failure, 297

truncated data values for migrated measures, 293

truncated decimal values for measures, 293

UNIX model creation problems, 299

using Check Model command, 262

tutorials

for new and experienced users, 15

two-digit years

handling, 57

Uunbalanced diagrams

definition, 392

unbalanced hierarchies

creating, 73


within a dimension, 73

union

of user classes, 222

unique moves

for categories in Transformer, 160


Index

uniqueness

avoiding problems with non-unique category

codes, 165

definition, 392

disabling verification to improve performance, 165

issues in alternate drill-down paths, 65

specifying, 287, 289

UNIX cubes

adding drill-through targets, 127

UNIX paths, 203

updating

authentication files, 295

authentication information in a model, 154

cube connections using cubeswap.bat, 157

cube metadata, 246

cubes incrementally, 256


incrementally, 161, 163

metadata, 142

user classes, 295

uploading

models to the server, 177

user class

definition, 392

user classes

enabling for a model, 153

setting up views, 152

union of views, 222

updating, 295

user class view

definition, 392

user class views

based on configured Series 7 namespace, 22

combining with dimension views, 156

customizing cubes, 144

setting up, 154

union, 222

user security

controlling access to metadata, 154

updating custom views, 154

Vvalue restriction subset definitions

definition, 392

values

blocked totals, 155

versions, 15

vertical axis

definition, 392

view

definition, 393

viewing

category actions in diagrams, 220

scope of a data source, 40

Wwait period settings

changing environment variable defaults, 299

warnings, 191, 198

weeks

partial, 81, 207

splitting, 81, 207

weighting

allocation by another measure, 207

What's this? help

in Transformer, 19

wildcard characters

troubleshooting ADO syntax errors, 285

wildcards, 219

YY2K, See Year 2000

Year 2000

specifying dates, 57

ymdinterval_exp

definition, 393

Zzeros

displaying, 107

measures, 291

zooming in and out

in Transformer diagrams, 22


Index