搜索
DATAWAREHOUSEDatawarehousingprovidesarchitecturesandtoolsforbusinessexecutivestosystematicallyorganize,understand,andusetheirdatatomakestrategicdecisions.Alargenumberoforganizationshavefoundthatdatawarehousesystemsarevaluabletoolsintoday'scompetitive,fastevolvingworld.Inthelastseveralyears,manyfirmshavespentmillionsofdollarsinbuildingenterprise-widedatawarehouses.Manypeoplefeelthatwithcompetitionmountingineveryindustry,datawarehousingisthelatestmust-havemarketingweapon——awaytokeepcustomersbylearningmoreabouttheirneeds.“So,youmayask,fullofintrigue,“whatexactlyisadatawarehouse?Datawarehouseshavebeendefinedinmanyways,makingitdifficulttoformulatearigorousdefinition.Looselyspeaking,adatawarehousereferstoadatabasethatismaintainedseparatelyfromanorganization'soperationaldatabases.Datawarehousesystemsallowfortheintegrationofavarietyofapplicationsystems.Theysupportinformationprocessingbyprovidingasolidplatformofconsolidated,historicaldataforanalysis.AccordingtoW.H.Inmon,aleadingarchitectintheconstructionofdatawarehousesystems,“adatawarehouseisasubject-oriented,integrated,time-variant,andnonvolatilecollectionofdatainsupportofmanagement'sdecisionmakingprocess.Thisshort,butcomprehensivedefinitionpresentsthemajorfeaturesofadatawarehouse.Thefourkeywords,subject-oriented,integrated,time-variant,andnonvolatile,distinguishdatawarehousesfromotherdatarepositorysystems,suchasrelationaldatabasesystems,transactionprocessingsystems,andfilesystems.Let'stakeacloserlookateachofthesekeyfeatures.(1).Subject-oriented:Adatawarehouseisorganizedaroundmajorsubjects,suchascustomer,vendor,product,andsales.Ratherthanconcentratingontheday-to-dayoperationsandtransactionprocessingofanorganization,adatawarehousefocusesonthemodelingandanalysisofdatafordecisionmakers.Hence,datawarehousestypicallyprovideasimpleandconciseviewaroundparticularsubjectissuesbyexcludingdatathatarenotusefulinthedecisionsupportprocess.(2)Integrated:Adatawarehouseisusuallyconstructedbyintegratingmultipleheterogeneoussources,suchasrelationaldatabases,flatfiles,andon-linetransactionrecords.Datacleaninganddataintegrationtechniquesareappliedtoensureconsistencyinnamingconventions,encodingstructures,attributemeasures,andsoon.(3).Time-variant:Dataarestoredtoprovideinformationfromahistoricalperspective(e.g.,thepast5-10years).Everykeystructureinthedatawarehousecontains,eitherimplicitlyorexplicitly,anelementoftime.(4)Nonvolatile:Adatawarehouseisalwaysaphysicallyseparatestoreofdatatransformedfromtheapplicationdatafoundintheoperationalenvironment.Duetothisseparation,adatawarehousedoesnotrequiretransactionprocessing,recovery,andconcurrencycontrolmechanisms.Itusuallyrequiresonlytwooperationsindataaccessing:initialloadingofdataandaccessofdata.Insum,adatawarehouseisasemanticallyconsistentdatastorethatservesasaphysicalimplementationofadecisionsupportdatamodelandstorestheinformationonwhichanenterpriseneedstomakestrategicdecisions.Adatawarehouseisalsooftenviewedasanarchitecture,constructedbyintegratingdatafrommultipleheterogeneoussourcestosupportstructuredand/oradhocqueries,analyticalreporting,anddecisionmaking.“OK,younowask,“what,then,isdatawarehousing?Basedontheabove,weviewdatawarehousingastheprocessofconstructingandusingdatawarehouses.Theconstructionofadatawarehouserequiresdataintegration,datacleaning,anddataconsolidation.Theutilizationofadatawarehouseoftennecessitatesacollectionofdecisionsupporttechnologies.Thisallows“knowledgeworkers(e.g.,managers,analysts,andexecutives)tousethewarehousetoquicklyandconvenientlyobtainanoverviewofthedata,andtomakesounddecisionsbasedoninformationinthewarehouse.Someauthorsusetheterm“datawarehousingtoreferonlytotheprocessofdatawarehouseconstruction,whilethetermwarehouseDBMSisusedtorefertothemanagementandutilizationofdatawarehouses.Wewillnotmakethisdistinctionhere.“Howareorganizationsusingtheinformationfromdatawarehouses?Manyorganizationsareusingthisinformationtosupportbusinessdecisionmakingactivities,including:(1)increasingcustomerfocus,whichincludestheanalysisofcustomerbuyingpatterns(suchasbuyingpreference,buyingtime,budgetcycles,andappetitesforspending),(2)repositioningproductsandmanagingproductportfoliosbycomparingtheperformanceofsalesbyquarter,byyear,andbygeographicregions,inordertofine-tuneproductionstrategies,(3)analyzingoperationsandlookingforsourcesofprofit,(4)managingthecustomerrelationships,makingenvironmentalcorrections,andmanagingthecostofcorporateassets.Datawarehousingisalsoveryusefulfromthepointofviewofheterogeneousdatabaseintegration.Manyorganizationstypicallycollectdiversekindsofdataandmaintainlargedatabasesfrommultiple,heterogeneous,autonomous,anddistributedinformationsources.Tointegratesuchdata,andprovideeasyandefficientaccesstoitishighlydesirable,yetchallenging.Muchefforthasbeenspentinthedatabaseindustryandresearchcommunitytowardsachievingthisgoal.Thetraditionaldatabaseapproachtoheterogeneousdatabaseintegrationistobuildwrappersandintegrators(ormediators)ontopofmultiple,heterogeneousdatabases.Avarietyofdatajoineranddatabladeproductsbelongtoth