© 2008 wellesley information services. all rights reserved. tips and tricks for using sap netweaver...

48
© 2008 Wellesley Information Services. All rights reserved. Tips and tricks for using SAP NetWeaver Business Intelligence 7.0 as your Enterprise Data Warehouse Dr. Bjarne Berg

Upload: gladys-paula-stephens

Post on 22-Dec-2015

218 views

Category:

Documents


3 download

TRANSCRIPT

© 2008 Wellesley Information Services. All rights reserved.

Tips and tricks for using SAP NetWeaver Business

Intelligence 7.0 as your Enterprise Data Warehouse

Dr. Bjarne Berg

2

In This Session ...

We will explore 6 large-scale EDW implementations, and see how to apply lessons to your strategy and projects.

Examine the difference between an evolutionary SAP NetWeaver BI data warehouse architecture and a top-driven design method.

Compare the results of using a data mart (bottom-up) approach to an EDW (top down) approach, and determine which approach best fits your requirements.

Explore the ways in which new SAP NetWeaver BI enhancements can support real-time data warehousing

We will look at common EDW pitfalls and how to leverage the SAP NetWeaver BI architecture in a large landscape using the Corporate Information Factory (CIF)

33

What We’ll Cover …

• Difference between evolutionary DW architecture and a design

• Data marts vs. Data warehouses

• Real-time Data warehousing

• The many mistakes of EDWs

• Successes and failures of six large-scale SAP BI-EDWs

• SAP NetWeaver BI architecture & Corporate Info. Factory (CIF)

• Wrap-up

4

Level of Pre-delivered ContentToolsets & accelerators

Analytical applications for specific industries

Lev

el o

f E

mb

edde

d A

naly

t ics

Complex (score cards, budgeting, planning, KPI)

Interactive Mgmt. reporting (OLAP, MQE)

Evolution of Data Warehousing

Emerging (1st generation)

Vertical approach (2nd generation)

Horizontal approach (2nd generation)

Integrated analytical (3rd generation)

Source: Mike Schroeck, David Zinn and Bjarne Berg, “Integrated Analytics – Getting Increased Value from Enterprise Resource Planning Systems”, Data Management Review, May, 2002;

Adapted: Bjarne Berg “How to Manage a BW Project”, BW & Portals Conference, 2007, Miami

Emerging (1st generation)

A General Conceptual Enterprise DW Architecture

Metadata

DataExtractionIntegration

andCleansingProcesses

Custom Developed Applications

DataMining

Statistical Programs

Query Access Tools

Data Resource Management and Quality Assurance

SummarizedData

SegmentedData Subsets

Functional Area

Summation

Marketingand Sales

Purchasing

CorporateInformation

Product Line

Location

PurchasingSystems

InvoicingSystems

GeneralLedger

External DataSources

Other InternalSystems

Translate

Attribute

Calculate

Derive

Summarize

Synchronize

Source Data ExtractOperationalData Store Transform

DataWarehouse BI Applications

Source: Bjarne Berg, “Introduction to Data Warehousing”,

Price Waterhouse Global System solution Center, 1997

SAP’s Technical EDW Architecture

Source: SAP AG

Visual Composer BI

Kit

UDI

XMLASAP

Query JDBC ODBO

Business Explorer Suite (BEx)

BEx Query Designer

Information Broadcasting

Web Analyzer

Web Application

Designer

MS Excel Add-in

Report Designer

BEx Web

DB Connect

BAPI Service API File XML/A

BEx AnalyzerBI Pattern

BI Consumer Services

BI Platform

Data Warehouse

Enterprise Portal

Analytic Engine Meta Data Mgr

KM

7

SAP’s EDW – Enablers - Query optimization

The SAP BI accelerator makes query response time 50-10,000 faster.

You use process chains to maintain the HPA engine after each data load

Both HP and IBM have standard solutions ranging from $32K to $200K+ that can be

installed and tested in as little as 2-4 weeks (+ SAP licensing costs)

SAP NW 2004s BI

HPA Engine/Adaptive Computing

DataAcquisition

InfoCubes

AnalyticEngine

SAP

BW

Any

tool

1. In-memory processing

2. Dictionary-based, smart compression using integers

3. High parallel data access / horizontal partitioning

4. Column-based data storage & access/vertical table decomposition

8

SAP’s EDW Enablers - Remodeling Tool Box

In older BW versions, if you forgot to include a field in your infocube, the rework was quite substantial and often involved reloading the infocube as well.

NW 2004s goes a long way to address the complaints that BW is a hard to maintain environment with ‘forever’ fixed models.

Source: SAP AG, Richard Brown, Aug. 2006

In NW2004s you get a new tool to add characteristics and key figures to your model.

9

SAP’s EDW Enablers - Central EDW Adm. & Tool reductions

SAP NetWeaver has solutions for a complete EDW architecture, including an Administrator

Cockpit for managing the system

In a custom data warehouse environment you need many tools:

- Data loads and transformations

- Scheduling of jobs

- Database management

- Data modeling

- Managed query environments

- On-line Analytical Processing tools (OLAP)

- Statistical analysis tools

- Data visualization tools

- Formatted reporting tools

- Web presentation tool

- Security administration tool

- EDW administration tool(s)

- Others ?

In a SAP data warehouse environment you need one tool:

SAP NetWeaver

10

SAP’s EDW Enablers - Global Tool Reach

The SAP Message: BO and SAP

provides

“Alignment, Extension &

Augmentation of two leading,

complementary BI & EIM solutions”

After the SAP’s Acquisition of Business Objects, many have questioned the long-term vision of SAP as the EDW. In Response, SAP published their tool integration vision in February 2008:

Source: SAP February 2008

11

SAP’s EDW Enablers - Long-term communicated vision

Notice that SAP Web Application

Designer is Replaced by Xcelsius+ in

2009 and a new tool called

‘Pioneer’ will be launched that

year also.

SAP has a long-term commitment to EDW and has published their 3-year tool plan so that customers can plan ahead.

Source: SAP February 2008

1212

What We’ll Cover …

• Difference between evolutionary DW architecture and a design

• Data marts vs. Data warehouses

• Real-time Data warehousing

• The many mistakes of EDWs

• Successes and failures of six large-scale SAP BI-EDWs

• SAP NetWeaver BI architecture & Corporate Info. Factory (CIF)

• Wrap-up

13

Design Vs. Evolution

An organization has two fundamental choices:

1. Build a new well architected EDW

2. Evolve the old EDW or reporting system

Both solutions are feasible, but organizations that selects an evolutionary approach should be self-aware and monitor undesirable add-ons and ‘workarounds”.

Failure to break with the past can be detrimental to an EDW’s long-term success…

14

ODS Vs. Data Warehouse Vs. Data Marts

To Understand the differences between DSO, Data Warehouses and Data Marts we can examine them in terms of usage, modeling and purpose:

• Specific application or workgroup focus

• Narrow scope

• Customized or stand alone analysis

• Interactive query

• Highly summarized

• Single subject and department oriented

• Uses dimensional data modeling

Data Store Objects (DSO) Data Warehouse Data Mart

• Acts as source to populate DW and marts

• Often used for operational reporting

• Detailed, atomic data

• Huge data volumes

• Integrated, clean data

• Cross-functional and cross-departmental

• Supports data mining

• May use denormalized form modeling (NOT dimensional)

• Provides mgmt reporting

• Summarized data

• Tuned to optimize query performance

• Multiple departments or processes

• May act as staging area for data marts

• Uses dimensional data modeling

15

Data Warehouse Vs. Data Marts - Implementation Sequence

There are several alternatives for an iterative approach to implementing the various storage structures, based upon organizational needs.

The various structures can be enterprise or departmentally focused.

They can be built first, middle, or last. They can be stand-alone or combined. The important point is to have a concept of the long term vision of the data warehouse project and how each type of structure is to be used.

A) ODS first: Start by building an enterprise data warehouse from a subject area perspective and then gradually move subsets of data to data marts. This approach may take a longer time to implement.

B) Data mart first: Start by building data marts to get data out to users quickly. This approach may encounter difficulties in integrating data from an enterprise perspective.

C) Data marts first within the framework or vision of an ODS: Start by developing a high-level enterprise or subject area data warehouse framework to guide the incremental development of the data marts or data warehouse.

16

Advantages of building the data marts first

There is a significant trend in the industry today toward building data marts first, then consolidating “backwards” to create the data warehouse and operational data store. There are several advantages to this approach:

A) Allows faster implementation

The average data mart may take 2-3 months to implement; the average EDW evolves over many iteration and may take years to mature. Several marts can be started in parallel.

B) Reduces political liability through alignment with a specific business need.

The mart can deliver value to the organization in a much shorter period of time and can focus on a specific business function or problem. The business sponsors will see faster results and can affirm their decisions with benefit analysis and feedback. This is important to maintaining interest and adequate funding levels for the program. This is in contrast to the time and complexity of building an enterprise data warehouse.

17

Advantages of building the data marts first (continued)

C) Limits risk while learning how to implement data warehouse.

Building very large databases of several Terabytes is inherently complex. Backup and recovery systems may require specialized hardware and software. Complex tuning may be necessary to achieve satisfactory query performance levels.

Identifying and defining data from many different sources creates opportunities for users and sponsoring departments to disagree. The ultimate business goals may be overshadowed by the technical and political difficulties of building the large warehouse. Starting small with a data mart, experimenting, and using the implementation as a learning experience, will reduce the risk and may actually result in a higher quality deliverable.

D) Costs less than an EDW. Initially, the economics of smaller scale hardware, software, and development staff

may contribute to lower costs for marts than EDWs.

18

Major Risks of building the Data Marts first

Data marts do not replace data warehouses.

The data mart is not the next step in data warehouse evolution. It must be planned and implemented as part of the overall architectural vision.

To be effective, you must maintain centralized control of data distribution to the mart in order to support the enterprise’s overarching warehouse goals of data quality, consolidation, and sharing.

Data marts also increase the complexity of the data warehouse environment with multiple extract, transform, and transfer routines.

There are some great risks of succumbing to political pressures.

Business units that demand a quick hit and a stovepipe implementation of data marts may only serve to undermine the best laid plans for an integrated and durable data warehousing program.

19

Risks of building the data marts first

If the IT department agrees to a bottom-up EDW, a strictly application specific approach, they may end up with multiple data marts that can not be integrated into a larger EDW/ODS view and which can not support analysis across different marts.

The bottom line is plan and build a reusable data and technical foundation (technology standards, data modeling principles, and integrated databases).

The Gartner Group estimates that resources required to manage a disjointed data mart environment are three times greater than an

integrated data warehouse architecture.

20

SAP’s Vision of Data Marts

If you insist on building data marts, you canalso use SAP’s newly acquired “Rapid Marts” tool from Business Objects.

Built with Data Integrator, SAP Rapid Marts are ready-made data marts for SAP. It has “pre-built data flows, business logic, and schema that understand the SAP meta-data”.

SAP Rapids Marts also include content that is immediately consumable by business users and can be deployed independent from an EDW implementation.

It supports data profiling and cleansing and can be “the first step toward a holistic EIM program or global EDW strategy”. In a prototype environment it can also provide early understanding of data quality problems. Source: SAP, Feb 2008

SAP has now inherited a tool for Data Marts that is

independent from the SAP NetWeaver Platform

2121

What We’ll Cover …

• Difference between evolutionary DW architecture and a design

• Data marts vs. Data warehouses

• Real-time Data warehousing

• The many mistakes of EDWs

• Successes and failures of six large-scale SAP BI-EDWs

• SAP NetWeaver BI architecture & Corporate Info. Factory (CIF)

• Wrap-up

22

Real-time SAP Enterprise Data warehousing gets better

NW 2004s has more features for updates that does not follow the typical asynchronomous (batch) updates. This include:

1. We can use XML to fill the PSA directly 2. Daemon-based update from delta queue (BW API)3. Daemon-based update of the ODS and minimal logging

Note: XML documents creates many tags that will slow down

large dataloads due to the size of each XML record (relatively large)

However, it works great for smaller streams of data.

23

Limitations of Real-time SAP Enterprise Data warehousing

There are some limitations depending on the version of SAP BI/BW you use. For versions 3.5 and higher, there are few limitations and they include:

You can only use real-time to load ODSs or PSA

A “normal” delta update and a real-time update cannot happen at the same time for the same DataSource and/or ODS

For data targets that subsequently store the real-time-supported ODS objects, real time data transfer cannot be used

InfoPackages that use real-time updates cannot be associated with InfoPackage Groups or Process Chains

Consider Using SAP Exchange Infrastructure (SAP-XI) to generate the XML documents from non-SAP Systems. This can help build a corporate data hub center

that can reduce the number of custom interfaces in the organization Tip

2424

What We’ll Cover …

• Difference between evolutionary DW architecture and a design

• Data marts vs. Data warehouses

• Real-time Data warehousing

• The many mistakes of EDWs

• Successes and failures of six large-scale SAP BI-EDWs

• SAP NetWeaver BI architecture & Corporate Info. Factory (CIF)

• Wrap-up

25

Common EDW Mistakes – Not Using Standard SAP Solutions

In the 1950s, you could buy a standard Sears house for $2,065 and pay $935 more to have it implemented on your own land

The customer’s who selected to buy the standard house were either “extremely happy” or “totally disappointed”.

When Sears examined why, they found a strong correlation between level of modifications to the home and unhappiness

You buy SAP NetWeaver for its pre-built content and connections to other SAP applications.

The more you add to the standard solutions, the harder it will become to realize the benefits you

sought in the first place.

26

Leveraging SAP Standard Content in The EDW

• As a guiding principle, map requirements to standard content before customizing

• However, you’ll probably also have external data sources that require custom ODSs and InfoCubes

• Customizing lower level objects will cause higher level standard objects to not work, unless you are willing to customize these also….

BW Content available (BI 7.0)BW Content available (BI 7.0)

• Cockpits ???• Workbooks 2,211• Queries 4,325• Roles 934• MultiProviders 402

• InfoCube 783• DSO objects 687• InfoObjects 14,368

36%

33%

31%

Mostly standard storage objectsSome customization

Highly customized storage objects

An example from a large manufacturing company

27

Billing

Number of billing documentsNumber biling line itemsBilled item quantityNet weightSubtotal 1Subtotal 2Subtotal 3Subtotal 4Subtotal 5Subtotal 6Subtotal ANet valueCostTax amountVolume

Customer

Sold-toShip-toBill-toPayerCustomer classCustomer group~ Customer country~ Customer region~ Customer postal code~ Customer industry code 1End user

Material

Material numberMaterial enteredMaterial groupItem categoryProduct hierarchyEAN/UPC

Time

Calendar yearCalendar monthCalendar weekCalendar day

Unit

Currency KeyUnit of MeasureBase unit of measureSales unit of measureVolume unit of measureWeight unit of measure

Billing information

Billing documentBilling itemBilling typeBilling categoryBilling dateCreation dateCancel indicatorOutput medium~ Batch billing indicatorDebit/credit reason codeBiling categoryReference documentPayment termsCancelled billing documentDivison for the order headerPricing procedure

Organization

Company codeDivisionDistribution channelSales organizationSales group

Logistics

PlantShipping/receiving point

Document details

Sales order document typeSales dealSales docuement

Accounting

Cost centerProfit centerControlling areaAccount assignment group

Personnel

Sales rep number

LEGEND

Delivered in standard extractorsDelivered in LO extractorNot in delivered Content -but in R-3

Standard content

How to Leverage Standard BI Content in the EDW

Storage Requirements

Storage Objects

Map functional requirements to the standard content before

you make enhancements

+

1. Create a model based on pre-delivered SAP BW content

2. Map your data requirements to the delivered content, and identify gaps

3. Identify where the data gaps are going to be sourced from

28

Common EDW Mistakes – No Tailored Approach

Each organization has different corporate cultures and considerations.

The Top-down approach is preferred in centralized organizations, and the bottom-up is preferred in decentralized organizations. Pick one approach and stick with it.

CO

NT

INU

E

TOP-DOWN APPROACH

Build a global data warehouse for the company, and proceed sourcing data from old legacy systems driven from a top-down approach.

CH

AN

GE

BOTTOM-UP APPROACH

Focus on a bottom-up approach where the BW project will prioritize supporting and delivering local BW solutions, thereby setting the actual establishment of the global Data Warehouse as secondary, BUT not forgotten.

29

Common EDW Mistakes – loose data standardsSome Many organizations place little value on enforcing data standards.

This include InfoObject, DSO and InfoCube naming standards. It also include naming conventions for queries and InfoAreas.

As a result, these organizations often have a ‘mess’ where it is hard to understand what is available without researching every field and data store.

It may also lead to problems integrating data with different data types and data lengths due to lack of enforcement

Develop your data standard and have an architect enforce them throughout the lifetime of the EDW.

Vchar2(15)

Char(18)Jims Query

AA Z0986 QueryX0C_K01 0SD_C03

0FIAR_O05 SUDHIRC99

30

Common EDW Mistakes – Lack of environment management

Some organization have a hard-time to say “No” to the business community.

As a result, their architecture often looks like mix-and-match of systems that was acquired to put out “urgent needs”.

In these organizations, multiple portals are common and overlapping reporting systems is the rule, not the exception.

EDWs are like marriages between IT and Business. You have to work at it constantly,

give it attention, and be faithful to the solution.

31

Common EDW Mistakes – lack of transport controls

Most companies have strong change management of their R/3 systems. However, it is common that the same organizations have very loose approval processes for their BI systems.

BI is becoming a mission critical system for most organizations and the same processes placed on the R/3 system should be applied to a production BI system.

Don’t allow quick-fixes and untested service packs and notes to be applied to the production box without adequate testing. BWQ is not for window dressing!!

If you want a stable BI system, you have to enforce testing and controls

32

Common EDW Mistakes – Poor Performance

When you build an enterprise data warehouse, you should plan for at least 10-15% of your project time for performance testing and tuning.

Click-stream analysis have shown the 50% of your casual audience will hit the refresh button or navigate away from your web site if the reports take more than 7 seconds.

If your query takes more then 20 seconds to run, you have major problems.

Get substantial amount of memory for caching and make sure your have a fast network and hardware resources.

#1 complaint of EDW is lack of performance. Consider BIA as part of your infrastructure

3333

What We’ll Cover …

• Difference between evolutionary DW architecture and a design

• Data marts vs. Data warehouses

• Real-time Data warehousing

• The many mistakes of EDWs

• Successes and failures of six large-scale SAP BI-EDWs

• SAP NetWeaver BI architecture & Corporate Info. Factory (CIF)

• Wrap-up

34

SAP EDW in 6 large Companies - Overview

In this EDW case study we are going to look at 6 diverse organizations and see their lessons learned in their own words

Organization Company 1 Company 2 Company 3 Company 4 Company 5 Company 6Industry Insurance Oil & Gas Oil & Gas Manufact. High-Tech Gov.

System BW 3.5 BI 7.0 BI 7.0 BI 7.0 BW 3.5 BI 7.0Number of Executive Users* 25 34 22 11 42 6Number of Casual users* 952 3,118 2,480 1,398 1,122 409Number of Power users* 34 14 46 23 89 7Number of non-SAP sources 6 4 11 3 13 24Number of SAP sources 31 107 86 24 144 9EDW data content (0-100%)** 80% 70% 75% 50% 50% 30%Lessons learned "Start with content

in finance and do few enhancements in the beginning"

"Have strong executive support

and think very long-term; 3-10 years"

"Spend serious time on end user

training and support. Sell the EDW internally"

"Shut-down competing reporting

systems; don't allow access databases"

"Users look at the query tools &

don't care about the EDW. Use

web tools"

"Data integration is 70% of the project.

Look at source systems early"

Overall satisfaction*** 7 8 8 8 7 9

Future Plans

BI Accelerator and web cockpits

Global rollout (Asia & Europe)

Global rollout (Europe)

Add new divisions in US &

purchasing

Rollout and add subsidiarie's

content

Rollout to the whole organization

* = actual users logged in within a 30 days period

** = estimated amount of organizational reporting done with EDW data

*** = Scale 1 to 9 (9 being highest and 5 being neutral)

35

SAP as the EDW in an Insurance Company

Go-live Year: 2003 (BW v. 3.0b)

Mistakes Made: Under estimated the time it would take to get the staff up to speed and trained in BW. Had no SAP web skills in-house and went with the wrong portal choice (non-SAP)

Successes: Built ‘foundation’ data stores first (AP, AR, GL, etc. before we started the individual department needs. This created a real EDW foundation instead of data marts. Now we are building more multiproviders and fewer new data stores. Because we built the EDW first, we can now deliver solutions faster.

Technology Challenges: Needed 3 app servers and more memory than first anticipated.Next Steps: Performance

tuning (BIA) and cockpits

Organization Company 1Industry Insurance

System BW 3.5Number of Executive Users* 25Number of Casual users* 952Number of Power users* 34Number of non-SAP sources 6Number of SAP sources 31EDW data content (0-100%)** 80%

Lessons learned "Start with content in finance and do few enhancements in the beginning"

Overall satisfaction*** 7

Future Plans

BI Accelerator and web cockpits

* = actual users logged in within a 30 days period

** = estimated % of org. reporting done with EDW data

*** = Scale 1 to 9 (9 being highest and 5 being neutral)

36

SAP as the EDW in Oil & Gas Company

Go-live Year: 2001 (BW v. 2.1c)

Mistakes Made: Stated with wrong area (MM). Should have done FI first and then HR. MM, APO and Motor Vehicle Fuel Tax reporting was too

complex and ambitious for the first implementation when we were learning.

Successes: Met budgets, deliverables and timelines. User satisfaction was very high when we went from only BEx workbooks to the web

templates. Upgrade to BI 7.0 was well received by developers and users.

Technology Challenges: Did not know how to performance tune the workbooks when we upgraded. They went from kilobytes to Megabytes. Needed on-line user training (CBT)

Next Steps: Adding the subsidiaries and corporate entities in Asia and Europe (650 more users)

Organization Company 2Industry Oil & Gas

System BI 7.0Number of Executive Users* 34Number of Casual users* 3,118Number of Power users* 14Number of non-SAP sources 4Number of SAP sources 107EDW data content (0-100%)** 70%

Lessons learned "Have strong executive support

and think very long-term; 3-10 years"

Overall satisfaction*** 8

Future Plans

Global rollout (Asia & Europe)

* = actual users logged in within a 30 days period

** = estimated % of org. reporting done with EDW data

*** = Scale 1 to 9 (9 being highest and 5 being neutral)

37

SAP as the EDW in another Oil & Gas Company

Go-live Year: 2000 (BW v. 2.0b)

Mistakes Made: No formal commitment to the EDW, that evolved over time (3 years). Did not have the top C-level commitment until 2003 and had to do a lot of rework to accommodate the new global vision.

Successes: We are 8 years into the EDW and it has been adapted as the core platform for global HR, finance and sales reporting. We have most divisions on the system and have retired six legacy reporting environments.

Technology Challenges: Needed more HW than originally planned. Performance was a real problem until 2006 when we started using the Broadcaster and cached some reports in memory.

Organization Company 3Industry Oil & Gas

System BI 7.0Number of Executive Users* 22Number of Casual users* 2,480Number of Power users* 46Number of non-SAP sources 11Number of SAP sources 86EDW data content (0-100%)** 75%

Lessons learned "Spend serious time on end user

training and support. Sell the EDW internally"

Overall satisfaction*** 8

Future Plans

Global rollout (Europe)

* = actual users logged in within a 30 days period

** = estimated % of org. reporting done with EDW data

*** = Scale 1 to 9 (9 being highest and 5 being neutral)

Next Steps: Adding European training and rollout (2 more R/3 systems)

38

SAP as the EDW in a Manufacturing Company

Go-live Year: 1999 (BW v. 1.2b)

Mistakes Made: Started too early with too ambitious scope. BW was not ready for EDW in 1999. Not until version 3.0b (2002) did we get a real ODS and could realize our earlier ideas of the EDW.

Successes: We kept the scope small and manageable, and had good consultants. The turnover rate on the project team has been low and the system was allowed to mature without business disruptions. We have consolidated three reporting groups into one and saved hundred of thousands of dollars in licenses each year.

Technology Challenges: Data integration was the hardest. We had to spend most of our project time on masterdata mapping & consolidation.

Organization Company 4Industry Manufact.

System BI 7.0Number of Executive Users* 11Number of Casual users* 1,398Number of Power users* 23Number of non-SAP sources 3Number of SAP sources 24EDW data content (0-100%)** 50%

Lessons learned "Shut-down competing reporting

systems; don't allow access databases"

Overall satisfaction*** 8

Future Plans

Add new divisions in US &

purchasing* = actual users logged in within a 30 days period

** = estimated % of org. reporting done with EDW data

*** = Scale 1 to 9 (9 being highest and 5 being neutral)

Next Steps: Add more functionality (purchasing) and rollout to purchasing group and the sales reps.

39

SAP as the EDW in a High-Tech Company

Go-live Year: 2003 (BW v. 3.1c)

Mistakes Made: User interface was not prioritized high enough. Executives and casual users hated BEx workbooks. We had to relauch the EDW in 2006 with a new web interface.

Successes: After the relaunch we have had success with user adaptation and have a functional steering committee and CFO sponsorship. Closing the financial books have gone from 5 days to 3.

Technology Challenges: Was unsure on how to interface our existing portal with SAP BI content (SSO). Security setup was hard and advise was too divergent. Process chains ran very slow until we tuned the ABAP.

Organization Company 5Industry High-Tech

System BW 3.5Number of Executive Users* 42Number of Casual users* 1,122Number of Power users* 89Number of non-SAP sources 13Number of SAP sources 144EDW data content (0-100%)** 50%

Lessons learned "Users look at the query tools &

don't care about the EDW. Use

web tools"

Overall satisfaction*** 7

Future Plans

Rollout and add subsidiarie's

content* = actual users logged in within a 30 days period

** = estimated % of org. reporting done with EDW data

*** = Scale 1 to 9 (9 being highest and 5 being neutral)

Next Steps: Add 2 more acquired companies to

SAP R/3 and BI.

40

SAP as the EDW in a Government Organization

Go-live Year: 2005 (BW v. 3.5)

Mistakes Made: Source data was in too many diverse old system with no real standards. We under

estimated the time it would take in integrate nine different mainframes, some that was 20+ years old. Should not used a ‘big-bang’ go-live.

Successes: Civilian and uniformed personnel worked well together and training was well received.

The data collection and reporting that used to take 14 days each month to produce, now takes 30 minutes.

Technology Challenges: During the BI 7.0 upgrade, the unicode conversion took long (did not

complete over the weekend). The BSP web templates had to be rebuilt completely.

Organization Company 6Industry Gov.

System BI 7.0Number of Executive Users* 6Number of Casual users* 409Number of Power users* 7Number of non-SAP sources 24Number of SAP sources 9EDW data content (0-100%)** 30%

Lessons learned "Data integration is 70% of the project.

Look at source systems early"

Overall satisfaction*** 9

Future Plans

Rollout to the whole organization

* = actual users logged in within a 30 days period

** = estimated % of org. reporting done with EDW data

*** = Scale 1 to 9 (9 being highest and 5 being neutral)

Next Steps: Add another maintenance organization

and work on web cockpits.

4141

What We’ll Cover …

• Difference between evolutionary DW architecture and a design

• Data marts vs. Data warehouses

• Real-time Data warehousing

• The many mistakes of EDWs

• Successes and failures of six large-scale SAP BI-EDWs

• SAP NetWeaver BI architecture & Corporate Info. Factory (CIF)

• Wrap-up

42

The Corporate Information Factory (CIF)

In 2001, Bill Inmon (the ‘father’ of DW) and Claudia Imhoff proposed a reporting architecture known as the CIF.

At the heart CIF’s reporting strategy is the EDW. It is the source of:

1. Decision Support System applications (APO, CRM, OLAP, Reporting etc).

2. Data Mining and APD

3. Departmental Data Marts

4. Access Media Accelerators (BIA)

Bill Inmon is a SAP BI technology advisor. He has

advised SAP on how to develop NetWeaver BI

43

Using the CIF – Reducing number of Platforms

NetWeaver helps by:

1. Reducing number of hardware servers

2. Consolidates the platform needs for budgeting, planning, forecasting and scheduling

3. Simplifies the platforms for web access, security, reporting and analysis.

FIInv Factory

DistWebOrder

….

SO

A / W

S

FIInv Factory

DistWebOrder

….

SO

A / W

S

2008

Other Enterprise Applications

DistributedApps

mySAPERP*1

FI/CO,HR

mySAP PLM*1

mySAPSRM*1

mySAP SCM*1

mySAPCRM*1

SAP NetWeaver

Portal Sec. EDW EnterprisePlatform

SimplifiedIntegration

TCO =

Enterprise Platform Cost

Cost of Applications

Cost of IntegratingApps & Platforms

+

+

End-to-EndService

Predictability

End-to-EndService

Predictability

A major CIF decision is how to integrate the solutions in as few platforms as possible.

CIF – provides a corporate framework for the EDW; NetWeaver provides the capabilities to do so with

one platformSolution

44

SAP’s Conceptual Enterprise Data Warehouse Architecture

Web Presentation/Portal/Mgmt Reporting

Mo

del

ing

an

d

Op

tim

izat

ion

Co

nso

lidat

ion

Bal

ance

d

Sco

reca

rd

Bu

dg

etP

lan

/Fo

reca

st

Sta

tuto

ryR

epo

rtin

g

Ad

Ho

c Q

uer

y an

d R

epo

rtin

g

DataMart DataMart DataMart DataMart

Data Warehouse

Integration Broker

ERP/CRM/SCM/External Sources

Bu

sin

ess

Pro

c. M

anag

emen

t

Co

nte

nt

Man

agem

ent

Kn

ow

led

ge

Man

agem

ent

Process Integration

SAP NetWeaver

People Integration

Information Integration

Source: SAP

SAP recognizes that we do not build EDWs, we are doing Enterprise Data warehousing. This is an on-going activity that merges information systems, people and processes.

EDW is an on-going activity

with continuous investment

needs.

4545

What We’ll Cover …

• Difference between evolutionary DW architecture and a design

• Data marts vs. Data warehouses

• Real-time Data warehousing

• The many mistakes of EDWs

• Successes and failures of six large-scale SAP BI-EDWs

• SAP NetWeaver BI architecture & Corporate Info. Factory (CIF)

• Wrap-up

46

COMERIT (Presentations, articles and accellerators)www.comerit.net

Enterprise Wide Data Warehousing with SAP BWhttps://www.sdn.sap.com/irj/sdn/go/portal/prtroot/docs/library/uuid/5586d290-0201-0010-b19e-a8b8b91207b8

Enterprise DataWarehousing – SAP Help http://help.sap.com/saphelp_nw70/helpdata/en/29/d9144236bcda2ce10000000a1550b0/frameset.htm

ResourcesResource

4747

7 Key Points to Take Home

• Plan Your Target EDW Architecture before you start the project.

• Enforce Standards and pick the right tools for the job

• SAP BI is no longer “leading” or “bleeding” edge and is used extensively as the EDW for large organizations

• If you are still on BI 3.5: Upgrade!

• SAP BI has many new tools that will enhance the front-end for end users. Your EDW will need them

• Critical to EDW success: reduce number of competing reporting system very quickly

• Hire an EDW Technical Architect if you have not already.

4848

Your Turn!

How to contact me:Dr. Bjarne Berg

[email protected]