cp-53: how to create effective data models to …...tuesday, 10:00-11:00 –gainesville 1-2 mehul...
TRANSCRIPT
CP-53:How to Create Effective Data Models to Support Your Dashboards and Reports
Keving Wang, Justin Sommer , Deltek
CONFIDENTIAL © Deltek, Inc. All Rights Reserved. 2
» Framework Manager Models - CQM vs. DQM – What it is and why it matters
» Data Module and Data Sets – Power your Dashboards without using Framework Manager
Session Overview
CER Role Capabilities
CONFIDENTIAL © Deltek, Inc. All Rights Reserved. 3
Capability Consumer CER
User
Advanced
CER User
CER
Developer
CER/BI
Admin
Web Admin
Run/Subscribe
ReportsX X X X X
Dashboard (View) X X X X
Dashboard (Create) X X X X
Report Authoring X X X
Data Module X X X
Event Studio X X X
Framework Manager X X
Admin Console X X
CONFIDENTIAL © Deltek, Inc. All Rights Reserved. 4
Relevant Roles for this Session» CER Developer & CER Admin
» IT
» Maybe ….. Advanced CER User
FM Models CQM vs DQMPerformance Improvement!
Compatible Query Mode (CQM)
»
CONFIDENTIAL © Deltek, Inc. All Rights Reserved. 6
CQM» Reports using the compatible query mode can benefit from new features implemented in
other components of the Cognos BI software stack, for example, RAVE visualizations. Do not
expect any enhancements to query generation or data processing for the compatible query
mode.
» CQM supports OLAP data sources such as Cognos Powerplay Cubes
»
CONFIDENTIAL © Deltek, Inc. All Rights Reserved. 7
Dynamic Query Mode (DQM)
DQM Details » Bursting reports: In CQM, when there is a master-detail relationship, CER will run the detail query for each row returned by the master query. This can result in the detail query being run a large number of time.
» Governor in FM named ‘(DQM) Master-Detail Optimization’: When this governor is set to ‘Cache Relational Detail Query’ then the DQM engine is able to process master-detail relationship more efficiently than they can be in CQM.
» 64 bit JDBC driver required (no OLE DB Driver needed)
» Java processes in CER 7.2.x (Cognos 11.0.13)
» Content manager & application server - the one java process you would see in Cognos 8.
» Query Service (XQE) to support Dynamic Query Mode - this was added in Cognos 10.
» Client container to support the new portal/dashboards/story boards/data modules/interactive
CONFIDENTIAL © Deltek, Inc. All Rights Reserved. 9
DQM Best Practices and Benefits• DQM is highly recommended when starting new projects (Deltek new
models are all DQM)
• Enabling DQM on existing projects is an optional migration (Deltek Legacy is still CQM) – Test all your reports!
• DQM performance may or may not be better
• Modifications may or may not be required for any report
• Complex reports might need modifications
• DQM enforces certain best practices that are optional in CQM
• DQM will not make certain assumptions that CQM does
• DQM may apply a different order of operations or join path resolution, which can potentially cause changes in a report output
• Deltek is getting different behaviors when switching existing models to DQM, make sure you do your testing before the switching.
CONFIDENTIAL © Deltek, Inc. All Rights Reserved. 10
DMR (Anyone? )
CONFIDENTIAL © Deltek, Inc. All Rights Reserved. 11
Data Module & Data Sets
Data Options
CONFIDENTIAL © Deltek, Inc. All Rights Reserved.13
Costpoint DB*
Packages
Data Module
Dashboards, Reports, Explorations (CER8)
Excel Upload*
Data Sets
CONFIDENTIAL © Deltek, Inc. All Rights Reserved. 14
Data SetsData sets can be used to gather a collection of items that you use frequently
Before We Get Started on the Data Module,Let’s Talk about…
» When defining a data set you
choose columns from a
package or data module…
» Apply filters to reduce the data
and summarize the data then…
» The data set is cached meaning
query performance is
improved
What Advantage Do Data Sets Give Me?
CONFIDENTIAL © Deltek, Inc. All Rights Reserved. 15
CONFIDENTIAL © Deltek, Inc. All Rights Reserved. 16
» Ctrl-Alt-M when
opening dataset
» Filter down data to
what you want to
analyze
» Schedule data set
refresh times
Modifying Datasets
Data Sets w/ Parquet
» Data sets use Parquet as columnar storage format
» Data set is stored in Content Store DB in compressed
and encoded to reduce the amount of storage; option to
store outside content store for growth management
» Parquet is one of the latest technologies used for
efficiency in data storage and scan/analytics, used
heavily in big data such as Hadoop and twitter; it is a
technology with improved read performance at the cost
of slower writes.
» It is a stable technology introduced in 2016 Oct.
» Excel upload in latest Cognos 11.1.x is loaded into
Parquet format as well
Data Sets Best Practice•Save frequently calculated expressions as columns.
This practice reduces the amount of expression evaluations at run time. Projecting, comparing, and
sorting simple column references and simple values (literals) is more efficient than evaluating
expressions.
•Avoid storing large numbers of columns that are never used by queries.
While data is both compressed and encoded to reduce the amount of storage, it’s still recommend to
avoid storing redundant or unnecessary columns.
•Use the sort option when creating a data set with column that is most frequently used in filters.
For large data sets, sorting the input can enhance the evaluation of predicates. Sorting the data on
the common column that is used in a filter, for example Country or Store, groups rows with the same
value. If a query includes predicates on that column, the query can determine more efficiently which
blocks of data it can ignore as it navigates the data.
CONFIDENTIAL © Deltek, Inc. All Rights Reserved. 19
Performance Testing with Data Sets
» Access via Browser (including Mobile)
» Easy to use for connecting different data sets together
» Map data from multiple sources
» Deltek Sources (Data Server currently on–premise only)
» Existing Framework Manager models
» Create calculations and more
» Uploaded Excel files (currently on premise only)
Data Module
» Combine data sets and Excel
sheets together with a data module
» Create the appropriate join
» Visualize the relations between
your different data objects
» And publish for others to use
Putting It All Together With Data Modules
CONFIDENTIAL © Deltek, Inc. All Rights Reserved. 21
CONFIDENTIAL © Deltek, Inc. All Rights Reserved. 22
» Data group on a numeric
column, by default ranges of
the groups will have equal
distribution
» Text column will allow you to
add in the data values to your
own groups
» Allows you to create buckets
that you can report on your
data like Quarters to Months
» Create rollups that do not exist
in Costpoint like Alt Proj
Data Groups in Data Modules
CONFIDENTIAL © Deltek, Inc. All Rights Reserved. 23
» Create your own drill
down path like PM to
Project
» Can leverage in reports
or dashboards
» Create multiple paths
across different data
fields
Navigation Paths
Navigation Path Setup
» Clean up blank values in your
data
» Create calculations so they can
be easily used on reports and
dashboards
» Note that Data Module features
will all be improved in CER 8
(CA 11.1)
Data Cleaning and Calculations in Data Modules
CONFIDENTIAL © Deltek, Inc. All Rights Reserved. 24
In CER 8- Let AI do its job!
CONFIDENTIAL © Deltek, Inc. All Rights Reserved. 25
Build an Awesome Dashboard in Seconds!
CONFIDENTIAL © Deltek, Inc. All Rights Reserved. 26
Data module and data sets» Future AI ability
» Better engine for query dataset, optional dedicated
query service
» The Query service advanced settings are used to
configure the Compute service that is used to process
data from uploaded files and data sets.
» The following Query service advanced settings can be
specified:
» qs.queryExecution.flintServer.maxHeap (Specifies the
maximum amount of memory that the Compute
service is allowed to use)
» qs.queryExecution.flintServer.sparkThreads
CONFIDENTIAL © Deltek, Inc. All Rights Reserved. 28
» Generate One Data set without security and set it to traverse only; Embed data module with
user security to filter dataset when user query the dataset; Schedule dataset refresh time by
trigger or time;
Data Module and Data Sets With Security
Notebook delivering a dataset
•Let users understand the underlying logic of “Code”. This is the opposite of how we usually think about
code. It is a series of “cells” containing executable code, or mark ups
•Notebooks can be saved and easily shared among users
•Underlying Python language has a standard library, and numerous third party libraries yielding a vast
array of existing code base and examples.
CONFIDENTIAL © Deltek, Inc. All Rights Reserved. 30
Less is Faster
Tuesday, 2:10 – 3:10 – Gainesville 1-2CP-52: Costpoint Business Intelligence: Upgrading from CER 7.1 to CER 7.2.3
Tuesday, 3:20-4:20 – Naples 1-2CP-53: Report Authoring Tips and Tricks in Costpoint Business Intelligence CER 7.2.3
Tuesday, 4:30-5:30 – Naples 1-2CP-55 Costpoint Business Intelligence CER 7.2.3- Executive Dashboards and Financial Reporting
Wednesday, 10:00-11;00 – Naples 3CP-56: As a Costpoint Business Intelligence CER End User, How Do I Leverage the New Features After We Upgrade?
Wednesday, 11:10-12:10 – Naples 3CP-53: Data Modeling-How to Create Effective Data Models to Support your Dashboards and Reports
Wednesday, 2:10 – 3:10 - Naples 3CP:54: Designing and Creating Information-Rich Interactive Dashboards in Costpoint Business Intelligence CER 7.2.3
Wednesday, 3:20 – 4:20 - Naples 1-2CP-50: Costpoint Business Intelligence Roadmap
Thursday 9:40-10:40 – Sarasota 1-2CP-59: Administration for Costpoint Business Intelligence CER 7.2.3 - Setting Up The New Security Model and How to Effectively Guard Your Data
Deltek BI Sessions
Tuesday, 10:00-11:00 – Gainesville 1-2 Mehul Desai, PCI
CP100: A Whole New World of Business Intelligence
Wednesday, 4:30-5:30 – Sarasota 1-2 – Hari Sury, Sury Analytics
CP-57:Analytics for GovCons - A Roadmap To Success
Thursday, 8:30-9:30 – Sarasota 1-2 – Rich Hardin, VSE
CP-51:Deltek Costpoint Business Intelligence (CER) Leveraging Analytics: The Old, The new And
Everything In Between
Customer/Partner BI Sessions