You are on page 1of 7

BIKnolwedgeSharewhitepaper

Cupertino,CA95014 info@erada.com

SAPBI Positioning theInfoCube inBI

SAPBIOptimizationseries InfoCubeguidelines


softwareengineeringIautomationISAPBI

HariGuleria,TeamBI
This whitepaper is provided by Erada, its copyright holders, and its contributors. This paper consists of confidential material and represents Erada open KnolwedgeShare material provided to customers as competencypapersandguidelinesexemplifyingourcompetitiveadvantageandthussubjecttocopyrightlawsofdistribution.ThispapermustnotbecopiedinpartorwholewithoutwrittenpermissionofEradaor itsauthor/s.Copyingofanypart/sorthewholepaperwillbedeemedasinfringementofcopyrightmaterialunderallapplicableUSlaws.ThewhitepapermaybedistributedAsIsandanyexpressorimplied warranties,including,butnotlimitedtowarrantiesofmerchantabilityandfitnessforaparticularpurposearedisclaimed.Innoeventshallthecopyrightowner,orcontributors,beliableforanydirect,indirect, incidental,special,exemplarilyorconsequentialdamageincluding,butlimitedto,lossofdata,orprofits;orbusinessinterruptions,,howevercausedandoronanytheoryofliability,whetherincontract,strict liabilityortort(includingnegligenceorotherwise)arisinginanywayoutoftheuseofthissoftware.SAPandBWareregisteredtrademarkofSAPGmbH,andSmartCube,InfoModeler,Datentechareregistered trademarkoferadaIncemail:hguleria@erada.com

Version: 8/17/2008

ResourceProfilefundamentalsResourceSkillProfile

Contents
Introduction Error! Bookmark not defined.

TheroleofInfoCubeinSAPBI KeytoBIreportingOptimizationInfoCube GettingclarityoftheroleoftheInfoCubeismostcriticalinBI Architecture and modeling. Differentiating that the ODS is thedatalayerandtheInfoCubethereportinglayerisintegral tosuccessinInformationdelivery

Key to BI reporting Optimization - InfoCube 2 Problem Statement Background information Introduction Management as the Lead InfoCube vs ODS/DSO ODS/DSO as a reporting object? InfoCube vs ODS/DSO indexing for BIA Conclusion 2 2 3 3 5 5 6 7

ThetopfourreasonsforfailureinInfoCubemodelingcanbe summarizedasfollows 1. Hasalwaysbeenamanualprocessthushasbecomemore ofanartthananarchitecture 2. CriticalC4impactanalysisisimpossibletoperformforany InfoCubemodeling 3. All InfoCubes have been modeled with a best guess modelingestimated,evenSAPBusinesscontentInfocubes 4. InfoCubemodelingiscomplexandtimeconsumingasitis ProblemStatement 1. 2. 3. InfoCubemodelingisextremelycomplextogetit entirelyright? Showustheskillprofileofasuccessfulandanot successfulBusinessIntelligenceArchitect? WhatarethefundamentalArchitectureCritical componentsthatcandecidethesuccessorfailure ofBIArchitecture?

1.

Backgroundinformation Thefundamentalbasisofthispaperisanagreement thatifallotherconsiderationsareconstantthenthe InfoCubeisamoreefficientInfoProviderthatthe ODS/DSOfromreporting. Manualmodelingisaverycomplextask,infactitis socomplexthatitcannotbedoneinanyoptimal manner EvenSAPdeliveredInfoCubesmaynotbetechnically optimizedandhaveapotentialforoptimizing

2.

3.

Justasanexample: A4dimension,7characteristicInfoCubehas47waysit canbemodeled. A13dimension,50characteristicInfoCubehasoverabillion waysitcanbemodeled. Manuallyconsideringabillionoptionsforanoptimaldesignis impossible. AutomationchecksmorethanabillionpossibilitiesagainstC4 checks,explainedlaterinthispaper,andprovidesanoptimal designveryrapidly.


ModelingforSAPBI: Knowhowtheplacement ofeachcharacteristicina dimensionwithimpactits C4attributes Developgoodproblem solvingskillsbylearning therelationshipsbetween dimensions,characteristics andtheirimpactonC4 attributes Solveskillsbylearningand alwaysemployingscientific andprovenmethodsof executionbasedonbest practicesandexperience. Accessteamstrengthsand weaknessesandleverage collectivestrengthforall tasks Visualizemodelingimpacts wherepossible Keepteamalways informedonallimpacts, failuresandsuccesses. Focustactical developmentson processes,Standardsand BIStrategicguidelines Develop&encourage managementskills PerformC4impactAnalysis

Introduction HowtoarchitectandmodelaSAPBIenvironmentshouldcreateconflictinany workplace,duetothesheerfactthatsofarithasbeenanartformandnotbasedon anystructuredrules. SAPhasprovidedODSandInfoCubeguidelines,howevertheseweredesignedfor kindergartenarchitectsandmodelersandarenotreallyrecommendedformatureBI organizations.Oneexampleistherecommendationthatallatomicdatamustbe locatedinODS/DSOandInfocubesmustonlybeusedforsummarizedreporting. However,maturearchitectsbothwithinSAPandoutsidewilleasilybeableto challengethisstatement.ThequestiontobeaskedisLeavingallotherfactorsaside whichInfoProviderisbetterforreportingInfoCubeorDSO?. SoasamatureorganizationwhatguidelinesshouldaBImanagerfollowformodeling Infocubes? Firstwehadleanmanufacturingthatturnedtheworldofmanufacturingintoa balancedandmeasurablemetricthatcouldbemeasured,tolerancesmappedand Manufacturingefficienciesreportedviaadvancedanalytics. NowweareheadingintothedomainofLeanBI. Mechanicalautomationishighlytechnicalandcomplex,andrequiresseveralyearsof trainingandworkexperiencetobecomeproficientinautomationtechnologies.Atthe sametimeautomationtechnologyalwaysimprovesperformanceandlowersoperational costs.Automationtechnologyisalsoconstantlyevolvingandencompassingmoreandmore processesandreplacingmanual,entrylevel,subjectivedecisionprocesseswithhighly integratedandmeasurablemetrics.Simpleautomationmaybecomplexbutcomplex automationcouldmakethingssimpler. BecausetraditionalmodelingofInfoCubesisamanualprocessweshallfindresistance fromartistsandcreativemanualmodelers.Howevermanagersandsubordinatessoon realizethattheinherentadvantagesoftheirpositions,andtheirworkquality,alignswith automationofcomplexmanualprocesses.Asasimilejustimaginebuildingamodern automobilewithlastcenturymanualprocesses.WhythencontinuewithmanualInfoCube modelingifthereisanautomationalternative. Hereareafewsuggestions.

ManagementastheLead Goodprocesses,standards,andgovernancemustcommencewithmanagement,whodrivethecreationofaCompanywide BICookbookconsistingofenterprisewidestandardizedprocessestobefollowedinalldevelopmentactivities.Thisdefines howeveryBIactivitymustbeperformed,i.e.allreportingmustbeprovidedfromMultiProviders.Thefundamental methodologiesmustnotbecomestatic,butevolvewithchangetoreacttodynamicchanges.Simultaneouslynoalternative mustbeallowedunlessthebookoffactsischanged. ThegoalofBIadministrationmustbetofirstlydefineandaligntheBIstrategytobusinessstrategygoals.Itmustdocument andcommunicatestrategicstandards,processesandmethodology,anditmustkeepthisdocumentdynamicbasedonnew functionalitiesandexpertexperience.AllrecommendationsmustbebasedonSAPbestPracticerecommendationsand guidelinesbasedonmultipleexpertfeedbackbackedbyprovencustomerexperiences.

3/7|P a g e

Managementmustworkwiththeirteamstomaximizeutilizationofavailableassets andencompassanyfacetofautomationwhetheritisintransformation,dataquality profilingorInfoCubemodeling.Automationalwaysleadstohigherefficiencies, predictableperformance,andloweredoperationalcosts. Goodprocessautomationmustpreconsidervariousoptions,reviewallalternatives andprovidethebestalternativeavailable.Inadditionitmustperformcriticaltests andassurethattheautomationprocessisinfactoptimizingtheprocesses. InfoCubeModeling

OneofthesetasksisInfoCubemodeling. InfoCubemodelingdemandsfourcriticalchecks,whichareneverpursuedduetolack oftime,budget,skillsorallthree. WecallthemtheC4checks.TherearethefundamentalrulesofInfoCubechecksin anymodelingorremodelingofanInfoCubemustcheckfor. C1basicallychecksInformationQualityimpact.ThebaselineforthisImpactanalysis startswithasIsreportingobjectsandreportpriority.Thischeckdoesnotcheckfor Dataquality.Thec1checkworksonabaseassumptionthatthedataqualityisa givenandthenproceedstoauditthatthereportingcapabilitiesoftheInfoCubeare notcompromised.Thatis,ifwehad12reportsbeingdeliveredfromtheInfoCubethis auditchecksthatthe12reportsarestilldeliverableandthatnoinformationcontent Technicalbackgroundisnecessary hasbeencompromised,neutralizedordeletedbymistake. torelatetotechnicaldetailsbut oftendoesnotrelatewellasa C2checksDataLoadimpacts.Thebaselineisthecurrenttimetakentoloaddatainto businessfacing,businessinteracting theInfoCube.AllsubsequentInfoCubemodelversionsmustbecomparedagainstthe personality.Technicalmanagers baselinetovisuallyreviewC2impact,i.e.theanticipatedimprovementor tendtobeinward/IT/technology degradationtodataloadperformanceintotheInfoCubebasedonthenewdesigns. focusedandnotneedfocused. Functionalbackgroundisnecessary C3checksQueryresponseimpact.Thebaselineforthecheckagainisthecurrent torelatetobusinesswithaclear systemloadonreportingfromtheInfoCube.AllsubsequentInfoCubemodelversions comprehensionoflimitationsand mustthenbecomparedforQueryRuntimeimpactforeachchangesuggested,i.e.the capabilitiesofSAPBI.Functional anticipatedimprovementordegradationofQueryperformance. managerstendtobebusiness focusedandmoreconcernedwith C4checksthefootprintoftheInfoCubeintheBI,thebaselineforthischeckisthe teamsandteamdynamics.They alwayspromptteammembersfor currentspaceoccupiedbytheInfoCube.AllsubsequentInfoCubemodelversions optimizingsolutions mustthenbecomparedforBIfootprintimpact,i.e.theamountofspacetheInfoCube occupiesintheSAPBIenvironment. ChooseTechnicalforITfacingand functionalforBusinessfulfillment. AgoodBWmanagerunderstandstheworktheyasktheirsubordinatestoexecute mustbetaskstheyarecapableofperformingandthattheypossesstherequisiteskills todotheassignedtasks.Thisknowledgeandcapabilityallowsthemanagerto monitorperformanceandprogressandtoavoidproblemsassociatedwiththehuman interpretationsandassociatederrorsinInfoCubemodeling. Agoodwaytoaccomplishthisistobuildindependentskillprofilesofeachsubordinateandthenoverlaythemintoa consolidatedskillmapofthewholeteamtovisuallyidentifybothindividualandteamskillprofile. Goodmodelingbeginswithgoodarchitecture.Agoodarchitecturemustencompasstechnology,DataQuality,Information, Businessunits,andorganizationarchitecturesallatthesametime.Thenproceedtoharmonizethemintoasinglecoherent BIArchitecture.

ManagingBIresources: AgoodBImanager Mustbebusinessfacingandhave excellentunderstandingof functionalities,capabilitiesand limitationsofdifferentaspectsof BI Mustmanageandnotgetinto handsdownproblemsolvingas thatdetractsthemfromtheir primarytaskofmanagement Mustunderstandstrengthsand weaknessesoftheircurrent models Mustknowthestrengthsand weaknessesofeachInfoAreato optimizeresourcemixand allocationsgearedtowards strategicsuccess Mustunderstandcriticalaspects businessneedsand functionalitiesthatcanmeet them

4/7|P a g e

UnderstandingInfocubes Manytypesofcubesformany reportingoptions Arenondestructiveobjects Datacannotbedestroyedonly changed.Changelogis maintained TruedataWarehouseereporting layer Leverageforreportingasfaras possible Usefordetailedandsummarized reporting Modelingisthekeytosuccess UnderstandingODS/DSO ManytypesofDSOformany reportingoptions Arenondestructiveobjects Datacanbedestroyedand changed.Changelogmaybe maintained TruedataWarehousedatalayer Canleverageforreporting Usefordetaileddataholding Canbereportingenabledornot NonreportingoDSloadfaster WriteoptimizedDSOneedtobe reloadedineachcycle

Goodmodelingissimplyutilizingallbestpracticesandpastexperiencetobuild modelsthatqualifytheC4impactoptimization. Theproblemwiththeabovestatementisthatgoodmodelingisnotcurrently possible.TomanuallyconductC4checksonewouldhavetobuildmany InfoCubes,loadeachoneofthemwithfulldata,buildqueriesineachofthe InfoCubes,andthenrunstatsoneachoneofthem.Thiscaneasilytake anywherefrom25to40daysperInfocube.Thisisnotacceptabletomost companies.Sotheendresultisasubjectivemodelthathasundergoneno comparativeimpactanalysis. ThisiswhereAutomationofInfoCubebecomesanimperativeoptiontowards BIsuccess. Automation EliminationofHumanerrors higherquality higher efficiency loweroperationalcosts=Progress ThisinturngivestheBIteammoretimetodohighervalueaddtasks. InfoCubevs.ODS/DSO TwoofthemostimportanttraitsofInfocubes,whencomparedtoODS/DSO, areindicesanddimensionalcapabilities. Indicesanddimensionalcapabilitiesareextremelyimportantindaytoday reportingbecausetheycollectivelyshortenthequerytimehorizonofselective reportingrequirementsinadynamicenvironmentofeverchangingreports andreportingneeds.Insimplertermsifthereare22regionsandyouneedto analyzeasingleregioninareporttheInfoCubedimensionandindiceswill onlyaccessthequerydimensionwhereastheODS/DSOquerywillneedto accessalltheregionsandthenselectivelyprovidetheregionrequestedduring runtime. MostSAPBIprofessional,bothinsideandoutsideSAP,wouldagreethatan InfoCubeisthepreferredreportingobjectbothforatomicandsummarized reporting.Thereshouldbefewsituations,fromaQueryperformancepointof view,whereanODS/DSOcouldeveroutperformanInfoCube,ifany.

ODS/DSOasareportingobject? SomeassumptionswhycompaniescontinuetouseODS/DSOasapreferredreportingInfoProvidercouldbe SAPsaidsoprobablythemostcommonreasonofall o Duringearlierdays,whenSapjustgotfamiliarwithODS,theyreleasedanODSmodelingwhitepaperthat recommendedthatduetotheODSbeingareportingobject o DuetothiswholeorganizationshavebeenwarehousingdetaileddatainflatfiledatabasesofODS,andonlytaking summarizeddataintoInfoCubes. o NumerousBIteamsfoundthedestructcapabilitiesofODS/DSOtotheiradvantagetoadapttotheeverychanging reportingneedsofbusinessusers.NotstrategicallypositioninganInfoCube,orMultiProviders,asthetrue reportinglayersintheSAPBIEDWenvironment. o SomecompaniescontinuetodelivermajorityofthereportsfromtheODS/DSOenvironment o AnalternateistoreviewwithSAPandselectBIexpertsthepros&consofInfoCubeandDSOreportingand finalizeastrategicreportinglayerthatconformstobestpracticesandoperationalefficiencies.Onemethod wouldbetotakethelargestandmostcomplexODSandconvertitintoanODSandcheckagainstalltechnical andbusinessneedstofinalizethedecision.ItwouldbenecessarytomodelaC4optimizedInfoCubetohaveany reasonabledecisionanalysistothisscenario 5/7|P a g e

DataloadtimeseeninlargeHighvolumelowvaluehighvolumeenterprises o AsInfocubestakelongertoloadsomemodelersreplacethereportingobject toaDSO,normallyawriteoptimizedDSO.Issuewithwriteoptimizedisthat itmustbeafullreloadpereveryloadcycle. o AnalternativecouldbetoreviewautomaticremodelingofDSOintoaC4 optimizedInfoCube,orremodelinganInfoCubetooptimalC4capabilities. ThereisahighpossibilitythatthedeltaInfoCubeloadtimemaybeshorter thatthefullDSOloadtime.Thisthenchangesthewholedynamicsofthe DSOvsInfoCubeasareportingobjectdecision Extremelyfastchangingreportingrequestsseeninlowvolumehighvalue enterprises o OnecustomeriscurrentlytakingallreportingobjectsfromInfocubesto ODS/DSO. o Thereasonprovidedisthesheerdynamicsofthedemandsofinformation users. o Thisisanexampleoflackofdocumentedprocesses,standardsandgovernance wherethechangefrequencycouldcompromisethebasicbestpracticesofthe EDWenvironment. o AnalternativeistoreviewtheArchitectureoftheBIenvironmentfromaself servicedesigncapability.InawelldesignedEDWenvironment,strategically modeledforselfservicethereshouldbemoreofasiloofverticalflowofdata. o Againstthisisthearchitecturethatisnotdrivenbydatawarehouse fundamentalsbutbyreportmartdesigns.Insucharchitecture,eitherdueto lackofaproperBIArchitecturestrategyorthesheerpressureofthebusiness ownerstodelivereveryreportyesterdaythedatawarehouseprinciplesbreak down. o AnalternativecouldbetoconductanArchitectureaudittowardsidentifying thedeltafromEDWbestpracticesandbestpracticeusageofInfoProviders.In suchcasesthereisahighpossibilitythatreportmartswillneedtobemerged intoSiloInfoCubesandmodeledinaccordancetoC4impactoptimization.In suchcasestheonlyoptionwouldbetoleverageautomatedInfoCube modelingtoolstorealignthearchitecturetowardsmeetingbusinessneeds. InfoCubevsODS/DSforBIAccelerator

Understanding BIAccelerator
IsdesignedforSAPBI WillencompassSAPapps IsnottheproverbialBlackBox appliance NeedsupportandRACI planning Has3distinctlayersof support

SupportingBIA

AsIsstatusofBIA OneofthemostcurrentBIAcceleratorstatementsisthattheBIAonlyaccepts InfoCubesforindexing.ThistranslatesintoafactthatonlytheInfoCubedatawillbe accelerated.WhichfurthertranslatesintoMurphyslawIfsomethinghastobreak,it will,inBIAcceleratorterminologythereportwillonlyrunasslowastheslowest InfoProvider,i.e.theODS/DSO.(alsorequestwhitepaperTheApplicationLayerinBIA Statedearlier MostSAPBIprofessional,bothinsideandoutsideSAP,wouldtendtoagreethatan InfoCubeisthepreferredreportingobjectbothforatomicandsummarizedreporting, andforindexingintotheBIA Translation IfyouhaveanyODS/DSOobjectassociatedwithyourreportthentheBIAcceleratorisaresourcethatiswastedduetothe facttheODS/DSOwillbecomeaQueryPerformanceDragtothereportinquestion.Thisissimilartospendingamillion dollarsoninstallingajetengineinyourracingcarandthenkeepingyoufootpressedfirmlyonthebrakesallthetime.Could eitherbeawasteofinvestmentornegativeutilizationofresourcesbothbadscenarios. Fact 1. InfoCubemustremainyoupreferredreportingInfoProviderofchoice,bothfordetailedandsummarizedreports. 6/7|P a g e

Layer1 Appliancelayer DonebyHWvendor Appliancesupport Storage/Files AllsoftwareotherthanBIA Allhardware Layer2 BIASWLayer DonebySAP SAPSupport BIASoftwaresupport Layer3 Applicationlayer Donebycompany Whichcubes WhichODS/DSOtoCubes IndexingCubestoBIA Datareconciliation Processchainmanagement LoadtoBIAAdmin BIAPlanning BIABladeforecasting BIAAlertmanagement BIASupportescalation

2. ConversionofODS/DSOtoInfoCubemustbealogicalnextsteptoBIAcceleratorinstallation ToBestatusofBIAcceleratorplannedforDec2009 SAPhasalreadyannouncedthattheBIAwillbeabletoindexdatafromODS/DSObuiltforBIA.Thisfunctionalityisplanned tobeavailableinDec2009. HoweverwewouldliketoclearlystatethattheBIAanalyticswillrunmoreefficientlyiftheypassthroughtheDimension modelingofanInfoCube Translation QueryresponsewillalwaysbebetterfromanInfoCubenomatterifyouarereportingfromtheBIandpassingthroughaBI Accelerator. ODS/DSOisaflatfileformatandforselectiveanalyticsthequeryhastopassthrougheveryrecord,i.e.aspecificRegional analyticinanODSreportthequerywilhavetorunthrougheveryregionnomatterhowyoumodelit.InandInfoCubeitwill onlyaccesstherespectiveregionifmodeledaccordingly. Fact InfoCubeswillalwaysremainthepreferredInfoProvidertoBIA,evenwhenthefunctionalityofODSisprovidedbySAP IfODS/DSOwillbeusedtoloadintoBIAduetoloadtimesconsiderAutomatedModelingforInfoCubesasaPOCtoreview changetostrategicInfoCubeutllization.

Conclusion
Ifanywhereyouseeanautomationoptiontoanymanualprocessembraceitreadily Automationleadsto EliminationofHumanErrors Higher,consistentandreliableQuality Higherefficiency Fasterturnaround LowerOperationalcosts PROGRESS RequestforademoofourAutomatedInfoCubemodelingoptionsforBIandseehowsimplewehavemadeoneofthemost criticalandcomplexTasksofBIModeling. Nooneexpectsperfectionfromamanager.Howeveramanagermustplantocontinuetoworksmarterandnotharder, mustcontinuetobebusinessfacingandnottechnologyfocused,mustcontinuetomeetstrategicbusinessneedsandnot accomplishtechnicalwonders. FinallythesweetnessofthepieisinBusinessUsersatisfaction.
AbouttheAuthor: HariGuleriaisaSrBIValueArchitect.HeisaBIBusinessValueStrategist&ArchitectatBIDataBridgeanderadaInclaunchingSAPBIsoftwareproductslikeSmartCube(automatedInfoCubemodeling),BIA Workbench(toolforBIAadministration)andRapidPro(ABAPoptimizationtool).Theteamiscurrentlyworkingona100%BIDataProfiling,ScoringandManagementtoolandaBOSmartUniverseduefor releasein2009 HariroutinelyworkswithcustomersastheirbusinessvalueArchitect,i.e.buildingsustainableBImethodologybyworkingwithbusinessownersandvendorpartners,assuringSAPandSAPBIBusinessValue enablement.Hariishighlybusinessneedfocused.HisstrengthisinBIStrategy,ArchitectureandBusinessValueassurance,andassistingBIcustomersrealizetheirbusinessgoals. PriortothisHariworkedatSAPAmericawiththeirValueRealizationgroupfocusingonSAPBIstrategyandvalueaudits.BeforethatHariheldnationalBI&EPpracticedirectorroleswithmajorSAPpremier partnercompanies.HaristartedSAPwithAndersenconsulting(nowAccenture)inSaudiArabiain1995. PriortoSAPHaricomeswithover9yearsofSr.sales,marketingandstrategicplanningexperiencewithEuropeanmultinationals.HehasbeenassistingcustomersmeettheirSAPR/3businessgoalssince1995 andBIgoalssince1997.HehasbeenaregularspeakeratASUG,andBITInationalconferences.Harihasover14fulllifecycleSAPprojectsunderhisbelt. Harimaybecontactedathguleria@erada.com


This whitepaper is provided by Erada, its copyright holders, and its contributors. This paper consists of confidential material and represents Erada open KnolwedgeShare material provided to customers as competency papers and guidelines exemplifyingourcompetitiveadvantageandthussubjecttocopyrightlawsofdistribution.ThispapermustnotbecopiedinpartorwholewithoutwrittenpermissionofEradaoritsauthor/s.Copyingofanypart/sorthewholepaperwillbedeemedas infringementofcopyrightmaterialunderallapplicableUSlaws.ThewhitepapermaybedistributedAsIsandanyexpressorimpliedwarranties,including,butnotlimitedtowarrantiesofmerchantabilityandfitnessforaparticularpurposeare disclaimed.Innoeventshallthecopyrightowner,orcontributors,beliableforanydirect,indirect,incidental,special,exemplarilyorconsequentialdamageincluding,butlimitedto,lossofdata,orprofits;orbusinessinterruptions,,howevercaused andoronanytheoryofliability,whetherincontract,strictliabilityortort(includingnegligenceorotherwise)arisinginanywayoutoftheuseofthissoftware.SAPandBWareregisteredtrademarkofSAPGmbH,andSmartCube,InfoModeler, DatentechareregisteredtrademarkofEradallcemail:info@bidatabridge.com

7/7|P a g e

You might also like