Upload
phyllis-burke
View
215
Download
1
Embed Size (px)
Citation preview
© 2004 Wellesley Information Services. All rights reserved.
Is your SAP BW ready for prime time?
- Lessons learned from real world go-lives
Dr. Bjarne BergLenoir-Rhyne College
2
User community
Is this your user community?
How can you avoid this BEFORE it happens?
3
What we will cover
• Evolutionary architectures (back and front-end)
• Design issues from a storage and performance perspective
• Complex architectures and lessons learned
• Testing of BW systems
We will examine systems from 4 real companies and discuss 5 more throughout this presentation
4
BW “post mortems”
• BW architectural issues at a global consumer products company
• A look at some BW design errors at the implementation level – a case study of a construction company
• More BW architectural issues – another angle of challenges through the eyes of a chemical company
• BW testing – what makes sense on a large BW project – a case study of a Fortune-500 manufacturing company
• Throughout this presentation we will also reference post implementation reviews at other companies. This include an oil and gas company, a European telecom company, a bank, an insurance company and an international materials joint venture organization.
Implementation
Architecture
Architecture
Testing
More
5
Architecture is Should be Based on Design
This architecture is a symptom of an evolution
This company started with a simple solution, but allowed the architecture
to take a back seat to “quick fixes”
SAP(Cost Center,
Product Costing,
Activity Based Costing)
PSA ODS
Flatfile1066B
1066BReports
ZABC
PC_C01
CCA_C11
ReportsCRD
More transforms
6
A complex solution to a reporting problem
Queries
Query views
Templates
Manual
Reports
Cubes
Lesson Learned: Sometimes non-SAP customization creates great
applications, but with poor flexibility and a high cost of ownership
Changing a query has major impact, when every query is published in a pre-run view and manipulated in custom Java code
Warning
7
What we will cover
• Evolutionary architectures (back and front-end)
• Design issues from a storage and performance perspective
• Complex architectures and lessons learned
• Testing of BW systems
We will examine systems from 4 real companies and discuss 5 more throughout this presentation
8
Number of BW storage objects
17
21
InfoCubes
Multicubes
ODS
Morrison HomesAvailable options ZOP_C01Jobcost details multicube ZMCJOBCOSJobcost history cube ZJC_C01Options sold ZOP_C02Pending close multicube (new) ZSDPSMC03Finance departmentMH AP line item cube ZAP_C01MH AR line item cube ZAR_C01MH general ledger cube ZARC01General ledger line item ODS ZGL_O01Purchasing/estimation departmentCost cube ZMM_C01MH property master (LOT) cube ZPM_C01Purchasing item data ZPU_C01Construction departmentInventory cube ZPS_C04Inventory multicube ZMCINVC01PS controlling MH ZPS_C01PS dates MH ZPS_C02PS controlling and dates cube ZPS_C03Sales departmentEarnest money estimation cube ZSD_C03MHSD overview ZSD_C01SD commissions cube ZSD_C02
InfoCubes in BW (as of 4/1/03)
An Implementation view
On the surface, this company appears to have a typical BW implementation with a set of InfoCubes, Multicubes and one ODS.
The implementation supported five different department and had been “live” for about one year.
9
Scope observations:The scope of the company's BW implementation was comprehensive and included 17 storage objects.
For a first time implementation in an organization unfamiliar with BW, this might be considered a somewhat ambitious scope.
In addition to this, the areas that were developed focused to a large extent on financial and project oriented detailed and integrated data. These are areas of a high degree of complexity.
A closer implementation look
10
Scope Recommendations
In general, a more phased approach might have been beneficial to the overall success of the BW rollout.
This scope management could have been accomplished by focusing on departments, or functional areas instead of a rollout of a large content area in a single phase.
The multiple phases (3-6 months each) might have been short in duration and still have accomplished to have all the content available earlier that what was achieved.
A closer implementation look
11
Report Status (per BW Team Lead)
Outstanding reports, 2
Reports with issues, 5
In production no issues,
15
What happened?
In this relative complex implementation users did not get the reports they asked for.
To meet the go-live date some reports went “live” with many issues, while other reports were not delivered as promised.
Lessons Learned: Don’t try to cram too much in your first go-live.
Consider multiple sequential, or parallel projects
12
Within typical design parametersApproaching recommended regular configuration parametersNot frequently used design parameters
Ledger
Infocubes Characteristics Navigational attributes
Time char-acteristics
Hierarchies Dimensions Key figures Record length
Complexity
Available options 20 4 4 2 9 4 Moderate ModerateJobcost history cube 57 0 1 0 12 1 High HighOptions sold 23 19 4 1 11 14 Low LowMH AP line item cube 52 13 4 3 10 15 High HighMH AR line item cube 29 0 4 1 8 20 Low LowMH general ledger cube 30 2 5 6 8 9 Low LowCost cube 14 0 3 2 8 2 Low LowMH property master (LOT) cube 24 2 4 1 11 7 Low LowPurchasing item data 13 10 5 3 11 15 Low LowInventory cube 33 4 2 3 9 23 Low LowPS controlling MH 44 37 5 10 12 35 Moderate HighPS dates MH 25 19 5 4 8 17 Low ModeratePS controlling and dates cube 34 48 3 5 9 30 Low HighEarnest money estimation cube 40 24 5 7 13 16 Moderate HighMHSD overview 49 28 5 7 13 27 High HighSD commissions cube 51 22 4 4 10 21 High High
BW InfoCubes Observations
Next: a look at the technical design
Just because an area is colored “red” does not mean it is wrong. However, a storage object with many red areas is worth taking a close look at.
Note
13
In general, a common BW configuration contains a set of characteristics that are used for analysis purposes. The number of these characteristics varies from implementation to implementations.
Typically configurations ranges from one to forty characteristics. The design depends largely on the requirements of the business, but there are some technical tradeoffs in load times when adding a very high number of characteristics. This is particularly true when these contains large text fields that are loaded at a high frequency and at a high number of records.
Infocubes Characteristics Navigational attributes
Time char-acteristics
Hierarchies Dimensions Key figures Record length
Complexity
Available options 20 4 4 2 9 4 Moderate ModerateJobcost history cube 57 0 1 0 12 1 High HighOptions sold 23 19 4 1 11 14 Low LowMH AP line item cube 52 13 4 3 10 15 High HighMH AR line item cube 29 0 4 1 8 20 Low LowMH general ledger cube 30 2 5 6 8 9 Low LowCost cube 14 0 3 2 8 2 Low LowMH property master (LOT) cube 24 2 4 1 11 7 Low LowPurchasing item data 13 10 5 3 11 15 Low LowInventory cube 33 4 2 3 9 23 Low LowPS controlling MH 44 37 5 10 12 35 Moderate HighPS dates MH 25 19 5 4 8 17 Low ModeratePS controlling and dates cube 34 48 3 5 9 30 Low HighEarnest money estimation cube 40 24 5 7 13 16 Moderate HighMHSD overview 49 28 5 7 13 27 High HighSD commissions cube 51 22 4 4 10 21 High High
BW InfoCubes Observations
CharacteristicsJust because an area is colored “red” does not mean it is wrong. However, a storage object with many red areas is worth taking a close look at.Note
14
Navigational attributes
Navigational attributes lends flexibility to the way users can access data. Common configuration consists of one to thirty attributes. While technically not incorrect one should review InfoCubes that does not contain any navigational attributes. One should also review of any Infocube that contained more than thirty. This may be an indicator that too much information is being placed in a single Infocube.
Hierarchies
Hierarchies are ways for users to "drill-down" into the data and is commonly used for analysis purposes. Typical configurations tends to have one to eight. One should review any Infocube with no hierarchy to validate this design with end user navigation, as well as question any design that contains a very high degree of these.
Technical stuff
Developer quote: “The consultants told me I could not cram all this stuff into one Infocube. I told them – you just watch me!!”
ClientIssue
15
Dimensions
BW allows for upto 13 dimensions to be created by a developer on a single Infocube. However, using all of these on a first implementation severely limits any future extensions without major redesign of the system (i.e. when adding more divisions to the system). One should review any InfoCubes that are approaching this limit.
Key figures
While no limitations are imposed by BW in terms of number of key figures (measures), typical implementations contains one to twenty of these. While a higher number of these may be required, there a are significant tradeoffs of load performance when a high number records are loaded (these are loaded with each transaction).
Dimensions
Lessons Learned: Don’t “paint” yourself in a corner on day one!!
16
Record length
In general, as the record length of an InfoSource increases, more data may be populated to the Infocube.
Since an Infocube might have more that one InfoSource, the length of each may be an indicator of the Infocube growth size as the company rolls out BW to other divisions.
You should review of the design of InfoSources with large record lengths to determine the true need of including all the fields in the Infocube vs. using alternate fields i.e. short text or codes, or removing them from the system..
Record Length
Lessons Learned: Don’t throw in the “kitchen sink” because it might come in handy one day…..
17
Data Loads
Infocubes Number ComplexityAvailable options 1 LowJobcost details multicube* 2 LowJobcost history cube 1 HighOptions sold 1 ModeratePending close multicube (new)* 2 LowMH AP line item cube N/A HighMH AR line item cube 1 ModerateMH general ledger cube 1 LowGeneral ledger line item ODS 2 HighCost cube 1 LowMH property master (LOT) cube 1 LowPurchasing item data 1 LowInventory cube 2 ModerateInventory multicube* 2 LowPS controlling MH 5 HighPS dates MH 1 ModeratePS controlling and dates cube 4 HighEarnest money estimation cube 1 ModerateMHSD overview 2 HighSD commissions cube 1 Moderate
BW Sources to InfoCubes/ODS and MultiCubesSources When fixing data load
problems, narrow the problem down quickly and focus on these areas.
Lessons Learned: Spend your time and effort in a
focused manner!!
18
Indexes BW index diagnosticStatistics BW diagnostic of statistics that is recommended to be updatedAggregates User designed aggregates (performance & existence)
Ledger
Performance
Morrison Homes Indexes Statistics AggregatesAvailable options Pass Pass NoneJobcost history cube Pass Yellow NoneOptions sold Pass Yellow NoneFinance department Indexes Statistics AggregatesMH AP line item cube Pass Yellow NoneMH AR line item cube Pass Yellow NoneMH general ledger cube Pass Yellow NonePurchasing/estimation dept. Indexes Statistics AggregatesCost cube Pass Pass NoneMH property master (LOT) cube Pass Yellow NonePurchasing item data Pass Pass NoneConstruction department Indexes Statistics AggregatesInventory cube Pass Yellow NonePS controlling MH Pass Pass NonePS dates MH Pass Pass NonePS controlling and dates cube Pass Pass NoneSales department Indexes Statistics AggregatesEarnest money estimation cube Pass Pass NoneMHSD overview Pass Pass NoneSD commissions cube Pass Yellow None
InfoCubes Performance observations (as of 4/1/03)
Before redesigning queries, take a look at the performance options of BW.
BW have many features that can be leveraged. This company did a somewhat mixed job in this area.
Lessons Learned: Cover the basics first before
you go “fancy”!!
19
High-Level tasksWeek
1Week
2Week
3Week
4Week
5Week
6Week
7
Copy BW existing system to new test environmentDevelop test casesExecute and document base test results from old environmentUpgrading the previous BW release on the new test environmentConduct regression testing of existing BW content (test cases)Conduct stress test of the existing queries and processes impactedAssemble a detailed assessment of impacts of upgrade Formal review session of upgrade test results and approval to progress.Backup of production environment and contingent recovery planningUpgrading of the production environment and testingFinal sign-off and approvals
How could they fix their system?
This company had major issues, but first it needed an upgrade to take advantage of the improved ODS (the company was still on version 2.0b)
Lessons Learned: Don’t be afraid to get better tools before you attack a problem..
20
Not everything has to be fixed…
InfoCube Tech name Redesign Level Suggested approach*Available options ZOP_C01 No N/AJobcost details multicube ZMCJOBCOS Yes High Redo (AP not in production today)Jobcost history cube ZJC_C01 Yes Moderate Reporting ODS Options sold ZOP_C02 No N/APending close multicube (new) ZSDPSMC03 Yes Low Redo based on new ZSD_C01 in ODS MH AP line item cube ZAP_C01 Yes Moderate To be devloped (not in production today)MH AR line item cube ZAR_C01 No N/AMH general ledger cube ZARC01 No N/AGeneral ledger line item ODS ZGL_O01 No N/ACost cube ZMM_C01 No N/AMH property master (LOT) cube ZPM_C01 No N/APurchasing item data ZPU_C01 No N/AInventory cube ZPS_C04 No N/AInventory multicube ZMCINVC01 No N/APS controlling MH ZPS_C01 Yes High Reporting ODS PS dates MH ZPS_C02 No N/APS controlling and dates cube ZPS_C03 Yes Moderate To be determinedEarnest money estimation cube ZSD_C03 Yes High Reporting ODS MHSD overview ZSD_C01 Yes High Reporting ODS SD commissions cube ZSD_C02 Yes Moderate To be determined
Leverage what you have and focus on the critical areas. Not everything has to be fixed
Lessons Learned: Don’t throw the “kids” out with the bath water!!!
21
What we will cover
• Evolutionary architectures (back and front-end)
• Design issues from a storage and performance perspective
• Complex architectures and lessons learned
• Testing of BW systems
We will examine systems from 4 real companies and discuss 5 more throughout this presentation
22
First, only one cube existed consisted of actual, plan and historical data.
• A MultiCube was built from 9 customized basic cubes.• 1 Infocube for Actuals• 4 InfoCubes for Current Plan Month for different versions• 4 InfoCubes for Plan Last Month for different versions.
• No reports ran against the basic cubes.
• All reporting was done via the multicube.
Integrated profitability at chemical company
GOTCHA!
23
SAP R/3
CE1DU01CE2DU02
Profitability Analysis
PSA
Reports
ZCOPA_V00
ZCOPA_V01
ZCOPA_V02
ZCOPA_V03
ZCOPA_V04
ZCOPA_V1H
ZCOPA_V2H
ZCOPA_V3H
ZCOPA_V4H
ZCOPA_MLT
FlatfilePlan & Actual
Integrated profitability at chemical company
This architecture is a result of “best intentions” and not much foresight.
An ODS might have served the company better than a large number of InfoCubes
with the same data
Lessons learned: “Everything in the world is a nail when you only
use a hammer”
This was intended to be a “scalable” solution
24
Integrated profitability at chemical company
Each of the nine cubes is large and have many objects and lots of daily data..
LowHigh102461373148ZCOPA_V3H
LowMedium102461373148ZCOPA_V04
LowHigh102461373148ZCOPA_V1H
LowHigh102461373148ZCOPA_V2H
LowMedium102461373148ZCOPA_V03
LowMedium102461373148ZCOPA_V02
LowHigh102461373148ZCOPA_V4H
LowMedium102461373148ZCOPA_V01
LowMedium102461373148ZCOPA_V00
ComplexityVolumeKey Figures
Unit Characteristics
Time Characteristics
DimensionsNavigational Attributes
CharacteristicsTechnical
LowHigh102461373148ZCOPA_V3H
LowMedium102461373148ZCOPA_V04
LowHigh102461373148ZCOPA_V1H
LowHigh102461373148ZCOPA_V2H
LowMedium102461373148ZCOPA_V03
LowMedium102461373148ZCOPA_V02
LowHigh102461373148ZCOPA_V4H
LowMedium102461373148ZCOPA_V01
LowMedium102461373148ZCOPA_V00
ComplexityVolumeKey Figures
Unit Characteristics
Time Characteristics
DimensionsNavigational Attributes
CharacteristicsTechnical
33413341ZCOPA_V04
6170680926ZCOPA_V03
614070629798ZCOPA_V02
725314747698ZCOPA_V01
Records Added
Records Transferred
Infocube
33413341ZCOPA_V04
6170680926ZCOPA_V03
614070629798ZCOPA_V02
725314747698ZCOPA_V01
Records Added
Records Transferred
Infocube
This architecture was partly intended to solve load problems, however BW version 3.x might have solved the problem through use of parallel loads
25
BW Security
• The security structure in place was not adequate, and was limited to only SAP logon restriction.
• The data was not restricted on any organizational level.
• Users could view any data present in the Cost and Profitability InfoCubes.
• There were approximately 100 users with access to corporate profitability
Since this company is publicly traded this is not only a design problem, it is also a possible SEC issue
26
Integrated profitability at chemical company
What Happened?Some queries was running longer than the older R/3-COPA environment, due partly to heavy use of currency and other calculations on the front-end.
Users were very unhappy !!!>46
>85
>231
>224
>203
>426
>158
Avg db (sec)
18Consolidated>192XXXXXX_06
620Detail>500XXXXXX_01
63Consolidated>317XXXXXX_03
24Consolidated>180XXXXXX_07
27Consolidated>253XXXXXX_05
20Consolidated>285XXXXXX_04
15Consolidated>470XXXXXX_02
No of output records
OutputAvg time (sec)
Query
>46
>85
>231
>224
>203
>426
>158
Avg db (sec)
18Consolidated>192XXXXXX_06
620Detail>500XXXXXX_01
63Consolidated>317XXXXXX_03
24Consolidated>180XXXXXX_07
27Consolidated>253XXXXXX_05
20Consolidated>285XXXXXX_04
15Consolidated>470XXXXXX_02
No of output records
OutputAvg time (sec)
Query
The fixes…..• Implement an ODS supported architecture• For query optimization, key figures can be pre-calculated during load
instead of created-on-the-fly on reports.• Leverage the OLAP cache for better response time during querying.• Aggregates can also improve reporting execution time.
293 queries – many ran in 3-8 minutes
27
What we will cover
• Evolutionary architectures (back and front-end)
• Design issues from a storage and performance perspective
• Complex architectures and lessons learned
• Testing of BW systems
28
A look at some testing a manufacturing company
• System Test
• Integration Test Cycle 1
• Integration Test Cycle 2 Load test Stress test Performance tuning
Q: How many testers does it take to change a light bulb?
A: Testers don't change anything. They just report that it's dark.
Q: How many testers does it take to change a light bulb?
A: Testers don't change anything. They just report that it's dark.
We assume that unit test is done during
developmentNote
29
BW Testing – The BIG PICTURE
Building Block
This company plans for adequate time between testing and go-live
30
Scope & Importance – a Manufacturing company
• Scope by Business Tracks The testing effort of the queries and reports
was organized around the Business Tracks
ProcurementDemand PlanningManufacturing ProcessPlanning & SchedulingDeliverOrderFinancials
In ScopeBusiness Track
ProcurementDemand PlanningManufacturing ProcessPlanning & SchedulingDeliverOrderFinancials
In ScopeBusiness Track
Since the company goes live with R/3 at the same time, this organization allows the BW teams to conduct integration tests with the process teams
31
Testing: Organization and Work
T e st an d Q u a lityM a na g er
T e st Co o rd in a to rs T e am Le a ds
O th e r R e sou rces(P ro ce ss T e a m s)
T e st P la n nin g Co o rd in a to rs
P ro ce ss M a na g e m e nt Test Strategy
Test Plan
Test Execution
Problem Resolution
Too often BW is treated as a “step child” to R/3 and real testing is executed “when possible”.
However, BW testing is a formal process and should have formally assigned resources and formally assigned tasks.
Don't Forget
A manufacturing company
32
System Test Cycle 2 - Milestones
• The BW system testing started behind the R/3 testing The BW teams developed queries and data stores while the
process teams conducted system test cycle-1. This allowed the BW team to refine requirements and to work
on better data in system test cycle-2.
Tasks\Dates December 2003 January 2004 February 2004 1-Mar 8-Mar 15-Mar 22-Mar 29-Mar 5-Apr
Identify People for Testing
Schedule Facilities
Prioritize Test Areas (Queries)
Send out Meeting Notice
Execute System Test
Document Results
Problem Resolution
112
2
3
45
67
8
9
10
11
Timing
"There's no time to stop for
gas, we're already late"
33
The Schedule….
• Each team had dedicated time in the test room.
• Food was provided (pastry, doughnuts etc).
• At least 2 testers (normally 3) was assigned to each query
• All test results was logged in a Lotus Notes database
3/1
3/2
3/3
3/4
3/5
3/6
3/7
3/8
3/9
3/1
0
3/1
1
3/1
2
3/1
3
3/1
4
3/1
5
3/1
6
3/1
7
3/1
8
3/1
9
3/2
0
3/2
1
3/2
2
3/2
3
3/2
4
3/2
5
3/2
6
3/2
7
3/2
8
3/2
9
3/3
0
3/3
1
4/1
4/2
DeliverCost and ProfitabilityOrderManufacturingPlan and schedulingDemand planningSource
Resolving outstanding
issues and re-testing
= Morning session 8:30 - noon= Evening session 12:30 - 5:00
Environment preparation
"If you didn't write it down,
it didn't happen."A manufacturing company
34
Test Progress Monitoring and Reporting
This company planned two types of meeting during system testing:
– Progress Meeting– Escalation Meeting
Progress MeetingWas held daily to monitor progress and resolve common issues – Business Analysts met with the development team. Other attendees included the test coordinator and the Test Problem Resolution (TPR) administrator,
Purpose: Discuss common issues, Monitor progress, Discuss any change in plans,
Duration: 30 minutesSolutionA manufacturing company
35
Escalation MeetingHeld weekly, or as required, to resolve open items that impacts multiple tracks
Other attendees include test coordinator, and the Test Problem Resolution (TPR) administrator,
Purpose: Discuss open issues, and take necessary action to speed up process, Resource allocation for escalated issues,
Duration: 30 minutes
Solution
Test Progress Monitoring and Reporting
A manufacturing company
36
Integration Test Cycle 2 – Load test execution
•Load test execution
Load for load test –20% of actual users (not named users) Schedule queries to run in background Execute each query while eCATT scripts are running Monitor BW system continuously Record findings in Lotus Notes database Meet daily with development team to resolve progress
Note
A manufacturing company
37
Integration Test Cycle 2 – Stress test execution
• Stress test execution Identify queries to be load tested Determine cutoff load for load test –40% of
actual users Schedule queries to run in background Execute each query while eCATT scripts are
running Monitor BW system continuously Record findings in Lotus Notes database Meet daily with development team to
resolve issues
Note
A manufacturing company
38
Integration Test Cycle 2 – Performance tuning
• Performance test execution Identify queries to be performance tuned Attempt tuning at query level Perform analysis based on benchmarks Build aggregates, indexes, etc if needed Record findings in Lotus Notes database Meet daily with development team to resolve issues
Keep this focused!!!Not all queries needs to run in
3 seconds..
39
Resources
• “Mastering the SAP Business Information Warehouse” Kevin McDonald et. al. chapter 10: “Performance Planning and Management” ISBN 0-471-21971-1
• ERP and Data Warehousing in Organizations: Issues and Challenges by Gerald G. Grant ISBN: 1-931-77749-7
• http://help.sap.com -- The SAP Help Portal. This is a great place to get access to the lastest and greatest from a documentation perspective.
40
• Spend some time thinking consciously about data architecture, and develop a formal “to-be” system diagram.
• Make all designs pass a review board, question overly complex designs or “work-arounds”.
• The tool has a high learning curve and training cannot substitute for experience. Plan on spending 10-15% of overall effort on performance tuning of queries and data loads.
7 Key Points to Take Home
Tip
41
• Do not attempt to “cram” all data into one cube. Keep InfoCubes logically organized and use multi-provider queries as needed.
• Do not succumb to using BW as a dumping ground, some reports belongs in R/3.
• Be very wary of evolutionary architectures. Be flexible, but create some “hard rules” that cannot be overridden by “quick fixes”.
• A line support organization should be established and be part of the development effort (knowledge transfer).
7 Key Points to Take Home
Warning