




已阅读5页,还剩69页未读, 继续免费阅读
版权说明:本文档由用户提供并上传,收益归属内容提供方,若内容存在侵权,请进行举报或认领
文档简介
DataWarehouse WhyDatawarehouse Themostcommonissuecompaniesfacewhenlookingatdataminingisthattheinformationisnotinoneplace Thebiggestchallengebusinessanalystsfaceinusingdataminingishowtoextract integrate cleanse andpreparedatatosolvetheirmostpressingbusinessproblems WhatisDataWarehouse Theideaofadatawarehouseistoputawiderangeofoperationaldatafrominternalandexternalsourcesintooneplacesoitcanbebetterutilizedbyexecutives lineofbusinessmanagersandotherbusinessanalysts Oncetheinformationisgathered OLAP on lineanalyticalprocessing softwarecomesintoplaybyprovidingthedesktopanalysistoolsforquerying manipulatingandreportingthedatafromthedatawarehouse DataWarehouseenvironment thesourcesystemsfromwhichdataisextractedthetoolsusedtoextractdataforloadingthedatawarehousethedatawarehousedatabaseitselfwherethedataisstoredthedesktopqueryandreportingtoolsusedfordecisionsupport DataWarehousingProcessOverview OperationalVs MultidimensionalViewOfSales CreatingADataWarehouse TheDataWarehouse TheDataWarehouseisanintegrated subject oriented time variant non volatiledatabasethatprovidessupportfordecisionmaking TheDataWarehouse IntegratedTheDataWarehouseisacentralized consolidateddatabasethatintegratesdataretrievedfromtheentireorganization Subject OrientedTheDataWarehousedataisarrangedandoptimizedtoprovideanswerstoquestionscomingfromdiversefunctionalareaswithinacompany TheDataWarehouse TimeVariantTheWarehousedatarepresenttheflowofdatathroughtime Itcanevencontainprojecteddata Non VolatileOncedataentertheDataWarehouse theyareneverremoved TheDataWarehouseisalwaysgrowing OperationalDatabasevs Datawarehouse OperationalDBSimilardatacanhavedifferentrepresentationsormeaningsFunctionalorprocessorientationCurrenttransactionFrequentupdating DataWarehouseUnifiedviewofalldataelementsSubjectorientationfordecisionsupportHistoricalinformationwithtimedimensionDataareaddedwithoutchange DataMart Adatamartisasmall single subjectdatawarehousesubsetthatprovidesdecisionsupporttoasmallgroupofpeople DataMart DataMartscanserveasatestvehicleforcompaniesexploringthepotentialbenefitsofDataWarehouses DataMartsaddresslocalordepartmentalproblems whileaDataWarehouseinvolvesacompany wideefforttosupportdecisionmakingatalllevelsintheorganization EnterpriseDataWarehouse EDW AlargescaredatawarehousethatisusedacrosstheenterprisefordecisionsupportEDWareusedtoprovidedataformanytypesofDSS includingCRM SCM BPM BAM PLM andKMS BPM BusinessperformancemanagementBAM BusinessactivitymonitoringPLM productlifecyclemanagementKMS Knowledgemanagementsystems Metadata Metadataisthedataaboutdata Inadatawarehouse metadatadescribethecontentsofadatawarehouseandthemannerofitsuseGoodmetadataisessentialtotheeffectiveoperationofadatawarehouseanditisusedindataacquisition collection datatransformation anddataaccess TheneedsforTechnicalmetadata Theuseofdatawarehousinganddecisionprocessingofteninvolvesawiderangeofdifferentproducts andcreatingandmaintainingthemetadatafortheseproductsistime consuminganderrorprone Automatingthemetadatamanagementprocessandenablingthesharingofthisso calledtechnicalmetadatabetweenproductscanreducebothcostsanderrors TheNeedsforBusinessmetadata Businessusersneedtohaveagoodunderstandingofwhatinformationexistsinadatawarehouse Theyneedtounderstandwhattheinformationmeansfromabusinessviewpoint howitwasderived fromwhatsourcesystemsitcomes whenitwascreated whatpre builtreportsandanalysesexistformanipulatingtheinformation andsoforth metadatainadatawarehouse Kimballliststhefollowingtypesofmetadatainadatawarehouse SourcesystemmetadataDatastagingmetadataDBMSmetadataRalphKimball TheDataWarehouseLifecycleToolkit Wiley 1998 ISBN0 471 25547 5 sourcesystemmetadata sourcespecifications suchasrepositories andsourcelogicalschemassourcedescriptiveinformation suchasownershipdescriptions updatefrequenciesandaccessmethodsprocessinformation suchasjobschedulesandextractioncode datastagingmetadata dataacquisitioninformation suchasdatatransmissionschedulingandresults andfileusagedimensiontablemanagement suchasdefinitionsofdimensions andsurrogatekeyassignmentstransformationandaggregation suchasdataenhancementandmapping DBMSloadscripts andaggregatedefinitionsaudit joblogsanddocumentation suchasdatalineagerecords datatransformlogs StarSchema Thestarschemaisadatamodelingtechniqueusedtomapmultidimensionaldecisionsupportintoarelationaldatabase Starschemasyieldaneasilyimplementedmodelformultidimensionaldataanalysiswhilestillpreservingtherelationalstructureoftheoperationaldatabase StarSchema FourComponents FactsDimensionsAttributesAttributehierarchies Figure13 14AThree DimensionalViewofSales Figure13 17AttributeHierarchiesinMultidimensionalAnalysis Facts NumericmeasurementsthatrepresentspecificbusinessaspectoractivityNormallystoredinfacttablethatiscenterofstarschemaFacttablecontainsfactslinkedthroughtheirdimensionsMetricsarefactscomputedatruntime Dimensions QualifyingcharacteristicsprovideadditionalperspectivestoagivenfactDecisionsupportdataalmostalwaysviewedinrelationtootherdataStudyfactsviadimensionsDimensionsstoredindimensiontables Attributes DimensionsprovidedescriptionsoffactsthroughtheirattributesNomathematicallimittothenumberofdimensionsUsetosearch filter andclassifyfactsSliceanddice focusonslicesofthedatacubformoredetailedanalysis AttributeHierarchies Providetop downdataorganizationTwopurpose AggregationDrill down roll updataanalysisDeterminehowthedataareextractedandrepresentedStoredinaDBMS sdatadictionaryUsedbyOLAPtooltoaccesswarehouseproperly StarSchema Astarschemaconsistsoffacttablesanddimensiontables Facttablescontainthequantitativeorfactualdataaboutabusiness theinformationbeingqueried Thisinformationisoftennumerical additivemeasurementsandcanconsistofmanycolumnsandmillionsorbillionsofrows Dimensiontablesareusuallysmallerandholddescriptivedatathatreflectsthedimensions orattributes ofabusiness Figure13 17StarSchemaForSales StarSchemaRepresentation Factsanddimensionsarenormallyrepresentedbyphysicaltablesinthedatawarehousedatabase Thefacttableisrelatedtoeachdimensiontableinamany to one M 1 relationship Factanddimensiontablesarerelatedbyforeignkeysandaresubjecttotheprimary foreignkeyconstraints Figure13 18OrdersStarSchema StarSchema Performance ImprovingTechniquesNormalizationofdimensionaltablesMultiplefacttablesrepresentingdifferentaggregationlevelsDenormalizationoffacttablesTablepartitioningandreplication Figure13 19NormalizedDimensionTables MultipleFactTables Practice Howtodesignastarschemaforanautoinsurancecompanytodoriskanalysis WhatistheObjective WhataretheFacts WhataretheDimensions WhataretheAttributes WhataretheAttributehierarchy AutoinsuranceDWstarschema DataWarehouseDesign GrainAdefinitionofthehighestlevelofdetailthatissupportedinadatawarehouseDrill downTheprocessofprobingbeyondasummarizedvaluetoinvestigateeachofthedetailtransactionsthatcomprisethesummary DataWarehouseImplementation TheDataWarehouseasanActiveDecisionSupportNetworkACompany WideEffortthatRequiresUserInvolvementandCommitmentatAllLevelsSatisfytheTrilogy Data Analysis andUsersApplyDatabaseDesignProcedures DataWarehouseImplementation ImplementingadatawarehouseisgenerallyamassiveeffortthatmustbeplannedandexecutedaccordingtoestablishedmethodsTherearemanyfacetstotheprojectlifecycle andnosinglepersoncanbeanexpertineacharea DataWarehouseImplementationRoadMap DataIntegrationandtheExtraction Transformation andLoad ETL Process Dataintegrationcomprisesthreemajorprocesses dataaccess theabilitytoaccessandextractdatafromanydatasource datafederation theintegrationofbusinessviewsacrossmultipledatastores andchangecapture theidentification capture anddeliveryofthechangesmadetoenterprisedatasources DataIntegrationandtheExtraction Transformation andLoad ETL Process Extraction transformation andload ETL Extraction readingdatafromadatabaseTransformation convertingtheextracteddatafromitspreviousformintotheformthatcanbeplacedintoadatawarehouseLoad puttingthedataintothedatawarehouse DataIntegrationandtheExtraction Transformation andLoad ETL Process DataCleanse Datacleansingordatascrubbingistheactofdetectingandcorrecting orremoving corruptorinaccuraterecordsfromarecordset table ordatabase Usedmainlyindatabases thetermreferstoidentifyingincomplete incorrect inaccurate irrelevantetc partsofthedataandthenreplacing modifyingordeletingthisdirtydata ETLtools AgoodETLtoolmustbeabletocommunicatewiththemanydifferentrelationaldatabasesandreadthevariousfileformatsusedthroughoutanorganization ETLtoolshavestartedtomigrateintoEnterpriseApplicationIntegration orevenEnterpriseServiceBus systemsthatnowcovermuchmorethanjusttheextraction transformationandloadingofdata ManyETLvendorsnowhavedataprofiling dataqualityandmetadatacapabilities On LineAnalyticalProcessing On LineAnalyticalProcessing OLAP isanadvanceddataanalysisenvironmentthatsupportsdecisionmaking businessmodeling andoperationsresearchactivities FourMainCharacteristicsofOLAPUsemultidimensionaldataanalysistechniques Provideadvanceddatabasesupport Provideeasy to useenduserinterfaces Supportclient serverarchitecture On LineAnalyticalProcessing AdditionalFunctionsofMultidimensionalDataAnalysisTechniquesAdvanceddatapresentationfunctionsAdvanceddataaggregation consolidation andclassificationfunctionsAdvancedcomputationalfunctionsAdvanceddatamodelingfunctions IntegrationOfOLAPWithASpreadsheetProgram Figure13 7OLAPServerArrangement SAP sBusinessInformationWarehouse anEnterprise WideInformationHub Anend to endenterprise wideinformationhubtosupportplanninganddecision making AcentraldatarepositoryofSAP non SAP current andhistoricalbusinesstransactionsandmetadata Timelyinformationtoalllevelsandroles fromanalysttoexecutive YearsofSAPfinancial logistic andhumanresourceinformationsystemsexperienceweddedwithmoderndatawarehousemethodologies ASampleOfCurrentDataWarehousingAndDataMiningVendors Table13 10 SuccessStoriesatPepsi Usingthedatawarehouse we vebeenabletoidentifyimportantitems findnationalsuppliersforthem andleveragethoserelationshipstoreducecosts Thankstothewarehouse Pepsicanmonitorpurchasingcomplianceattheuserlevel anabilitythathasboostedpriceandproductcompliancewellover90percent Thewarehousealsohelpsensure100percentsalestaxcompliance saysBridgman Sincegoingonlinein1995 thewarehousehashelpedgenerateprocurementsavingsinexcessof 100million LevelsofDWSupportforEnterpriseDecisionMaking Theneedforreal timedata AbusinessoftencannotaffordtowaitawholedayforitsoperationaldatatoloadintothedatawarehouseforanalysisProvidesincrementalreal timedatashowingeverystatechangeandalmostanalogouspatternsovertimeMaintainingmetadatainsyncispossibleLesscostlytodevelop maintain andsecureonehugedatawarehousesothatdataarecentralizedforBI BAtoolsAnEAIwithreal timedatacollectioncanreduceoreliminatethenightlybatchprocesses Real Time ActiveDataWarehouse RDW ADW Loadingandandprovidingdataviathedatawarehouseastheybecomeavailable ExpandtraditionaldatawarehousefunctionsintotherealmoftacticaldecisionmakingEmpowerdecisionmakingwheninteractdirectlywithcustomersandsuppliers Real TimeDataWarehousing DataWarehouseAdministration Duetoitshugesizeanditsintrinsicnat
温馨提示
- 1. 本站所有资源如无特殊说明,都需要本地电脑安装OFFICE2007和PDF阅读器。图纸软件为CAD,CAXA,PROE,UG,SolidWorks等.压缩文件请下载最新的WinRAR软件解压。
- 2. 本站的文档不包含任何第三方提供的附件图纸等,如果需要附件,请联系上传者。文件的所有权益归上传用户所有。
- 3. 本站RAR压缩包中若带图纸,网页内容里面会有图纸预览,若没有图纸预览就没有图纸。
- 4. 未经权益所有人同意不得将文件中的内容挪作商业或盈利用途。
- 5. 人人文库网仅提供信息存储空间,仅对用户上传内容的表现方式做保护处理,对用户上传分享的文档内容本身不做任何修改或编辑,并不能对任何下载内容负责。
- 6. 下载文件中如有侵权或不适当内容,请与我们联系,我们立即纠正。
- 7. 本站不保证下载资源的准确性、安全性和完整性, 同时也不承担用户因使用这些下载资源对自己和他人造成任何形式的伤害或损失。
评论
0/150
提交评论