《信息科學(xué)類專業(yè)英語》課件第10章

Lesson10DataWarehouseOverview

(第十課數(shù)據(jù)倉(cāng)庫(kù)概論)

Vocabulary(詞匯)ImportantSentences(重點(diǎn)句)QuestionsandAnswers(問答)Problems(問題)

TheworddatawarehousewasfirstdevelopedbyBillInmonintheearly1990s.Hereferredtoitasbeingaintegratedcollectionofinformationthatcouldhelpcompaniesandorganizationsmakebetterdecisions.

Tobeeffective,adatawarehousehadtobeintegrated,subjectoriented,non-volatile,andtimevariant.Inthisarticle,Iwillgooverallthesefactorsindetail.Ifyouarebuildingadatawarehouse,itisimportantforyoutounderstandwhytheyareimportant.

Beingsubjectorientedmeansthatthedatawillprovideinformationaboutaspecificsubjectratherthantheinformationaboutthefunctionsofacompany.Becauseadatawarehouseissubjectoriented,itwillallowyoutoanalyzeinformationthatisconnectedtoaspecificsubject.Beingintegratedmeansthatthedatathatiscollectedwithinthedatawarehousecancomefromdifferentsources,butcanbecombinedintooneunitthatisrelevantandlogical.Havingatime-variantmeansthatalltheinformationwithinthedatawarehousecanbefoundwithagivenperiodoftime.[1]

Itisimportantthattheinformationcontainedwithinadatawarehouseisstable.Whiledatacanbeadded,itshouldneverbedeleted.Thispropertyisreferredtoasbeingnon-volatile.Whenacompanyusesadatawarehousethatisstable,thiswillallowthemtogetabetterunderstandingoftheoperationswithintheircompany.Despitethefactthatthesetermswerefirstcoinedinthe1990s,theyarestillhighlyaccuratetoday.However,itshouldbenotedthatsomedatawarehousesarevolatile.Thereasonforthisisbecausemanymoderndatawarehousesdealwithterabytesofdata.Becausetheymuststoreterabytesofdata,manycompaniesareforcedtodeletesomeoftheirinformationafteracertainperiodoftime.Forinstance,somecompanieswillsystematicallydeletedatathathasreachedthreeyearsofage.Beforeadatawarehousecanbebuilt,thecorrectdatamustbelocated.Generally,theinformationthatwillbeaddedtothewarehousewillcomefromdailyinformationorhistoricalinformation.Thehistoricalinformationmaybestoredinalegacysystem,andischallengingtoextract.

Thedesignofthedatawarehouseisimportantaswell.Itisimportantfordesignerstomakesurethedesignisconsistentwiththequeriesthatwillbeconductedwithinthewarehouse.Todothissuccessfully,itisimportantfordesignerstounderstandthedatabaseschema.Itiscrucialtomakesurethedatawarehouseisdesignedcorrectly,asitisdifficulttorecreatesomeformsofdata.Anotherimportantaspectofdatawarehousesisdataacquisition.Dataacquisitioncanbedefinedastransferringdatafromasourcetothewarehouse.Dataacquisitionisoneofthemostexpensivepartsofbuildingadatawarehouse.ThisprocesswilloftenbeconductedwithanETL(Extracting,TranslatingandLoading)tool.

Asofthistime,therearejustover50ETLtoolsbeingsold.Itmaycostacompanymillionsofdollarsinordertotransferdatafromsourcestothewarehouse.Oncetheinitialdatahasbeentransferredtothedatawarehouse,theprocessmustberepeatedconsistently.Dataacquisitionisacontinuousprocess,andthegoalofacompanyistomakesurethewarehouseisupdatedonaregularbasis.Whenthewarehouseisupdated,itisoftenhardtodeterminewhichinformationinthesourcehaschangedsincethepreviousupdate.Theprocessofdealingwiththisissueiscalledchangeddatacapture.Thisprocesshasbecomeaseparatefield,andthereareanumberofproductscurrentlybesoldtodealwithit.

Itisimportantfordatatobecleanedbeforeitcanbeplacedinthewarehouse.Thedatacleansingprocessisusuallydoneduringthedataacquisitionphase.Anydatathatisplacedinawarehousebeforebeingcleanwillposeadangertothesystem,anditcannotbeused.Thereasonforthisisbecausethedatamaynotbecorrectifitisnotcleaned,andacompanymaymakeincorrectdecisionsbasedonit.Thiscouldleadtoanumberofproblems.Forexample,alltheinformationwithinadatawarehousethatmeansthesamethingmustbestoredinthesameform.Ifthereisinformationthatreads“MS”and“Microsoft”,eventhoughtheymeanthesamething,onlyoneofthemcanbeusedtorecognizetheelementwithinthedatawarehouse.1DataWarehouseTools

Thereareanumberofimportanttoolswhichareconnectedtodatawarehouses,andoneoftheseisdataaggregation.Adatawarehousecanbedesignedtostoreinformationbasedonacertainlevelofdetail.Forexample,youcanstoredatabasedoneachtransaction,oryoucanstoreitbasedonasummary.Theseareexamplesofdataaggregation.Whendataissummarized,thequerieswillmoveatamuchfasterrate.However,someoftheinformationmaybelostduringaquery,andthisinformationmaybeimportantforsolvingacertainproblem.

Beforeyoudecidewhichoneyouwilluse,itisimportanttoweighyouroptionscarefully.Onceyouhavecarriedoutanoperation,youwillneedtorebuildthewarehouseinordertoundoit.Thebestwaytohandlethissituationistomakesurethedatawarehouseisconstructedwithalargeamountofdetail.However,thecostforthiscanbehugedependingonthestorageoptionsyouchoose.Onceyouhavefilledyourdatawarehousewithimportantinformation,youwillwanttousethisdatatohelpyoumakesmartinvestmentdecisions.Thetoolsthatcanallowyoutodothiswillfallunderatopicthatiscalledbusinessintelligence.

Businessintelligenceisafieldwhichisverydiverse.ItiscomprisedofthingssuchasExecutiveInformationSystems,DecisionSupportSystems,andBusinessintelligencecanfurtherbebrokendownintoafieldthatiscalledmulti-dimensionalanalysistools.Thesearetoolsthatwillallowausertoviewdatafromawidevarietyofangles.AquerytoolwillallowausertosendSQLquerieswithinawarehousetolookforresults.Dataminingisalsoafieldthatfallsunderbusinessintelligence,andwillallowyoutolookforpatternsandrelationshipswithinadatawarehouse.

Anothertoolthatisconnectedtodatawarehousesisdatavisualization.Thetoolsthatareusedfordatavisualizationwillpresentvisualmodelsofdata.Thisdatacouldcomeintheformofintricate3Dimages.Thegoalofdatavisualizationistoallowtheusertoviewtrendsinamethodwhichiseasiertounderstandthancomplicatedmodelsthatarebasedoffstatistics.OnetoolthatisallowingthisfieldtoadvanceisVRML,orVirtualRealityModelingLanguage.Inorderfordatawarehousestofunctionproperly,itisalsoimportanttoplaceanemphasisonmetadatamanagement.Metadatacanbedescribedasbeing“informationaboutinformation”.

Metadatamustbemanagedwhendataisacquiredoranalyzed.Metadatawillbeheldinarepository,andcangiveyouimportantinformationaboutmanyofthedatawarehousetools.Theprocessofproperlymanagingmetadatahasbecomeasciencewithinitself.Ifitisdoneproperly,thecompanycangreatlybenefit.Thereasonwhyitisimportantisbecauseitcanalloworganizationstoanalyzethechangesthatoccurwithindatabasetables.Thisisatoolthatplaysanimportantpartoftheconstructionofadatawarehouse.

Datawarehousingisafieldwhichissomewhatcomplicated.Therearemanyvendorswhoareattemptingtoadvertisethetools,butthecostandcomplexityinvolvedwiththeproductshasnotallowedthemtobeusedbyalargenumberofcompanies.Anycompanythatisthinkingofusingdatawarehousesmustmakesuretheyhavetakenthetimetoreviewandunderstandthetechnology.Itcanonlybeusefulifyouknowhowtouseit.Onceyouunderstandandacquirethetechnology,itispossibleforyoutogainapowerfuladvantageoveryourcompetitors.Thishasmadedatawarehousesattractivetomanycompanies.

Oneofthebiggestadvantagestodatawarehousesisthattheyallowyoutostoreinformationthatyoucanusetoimprovethemarketingstrategiesofyourcompany.Notonlycanyouimprovethemarketingstrategies,butyouwillalsobeabletomakestrategicdecisionsbasedontheinformationyouhavecompiledandorganized.Withtechniquessuchasdatamininganddatavisualization,youwillbeabletodiscoverimportantpatternsthatyoudidn’tknowexisted.Thepatternsthatyoudiscovercanallowyourcompanytoearnlargeprofits.2DataWarehousingMethods

Mostorganizationsagreethatdatawarehousesareausefultool.Theybenefitfromtheabilitytostoreandanalyzedata,andthiscanallowthemtomakesoundbusinessdecisions.Itisalsoimportantforthemtomakesurethecorrectinformationispublished,anditshouldbeeasytoaccessbythepeoplewhoareresponsibleformakingdecisions.

Therearetwoelementsthatmakeupthedatawarehouseenvironment,andthesearepresentationandstaging.Thestagingcouldalsobeknownastheacquisitionarea.ItiscomposedofETLoperations,andoncethedatahasbeenprepared,itwillbesenttothepresentationarea.

Whenthedataisplacedwithinthepresentationarea,anumberofprogramswillanalyzeandreviewit.Whilemanyorganizationsagreeontheoverallgoalofdatawarehouses,theapproachestobuildingthemmaydiffer.Attemptingtousedatamartsaloneisnotagoodapproach,becausetheyaregearedtowardsdepartments.Inadditiontothis,attemptingtousedatamartsalonewillbeinefficient,andyouwillrunintoanumberoflongtermproblems.Therearetwotechniquesforbuildingdatawarehousesthathavebecomeverypopular.ThesearetheKimballBusArchitectureandtheCorporateInformationFactory.

WiththeKimballtechnique,theroughdatawillbetransformedandrefinedwithinthestagingarea.Itisimportanttomakesurethedataisproperlyhandledduringthisstep.Duringthestagingprocess,theroughdatawillbepulledfromthesourcesystems.Whilesomeofthestagingprocessesmaybecentralized,otherswillbedistributed.Thepresentationareawillhaveadimensionalstructure,andthismodelwillholdthesameinformationasastandardmodel.However,itwillbeeasiertouse,anditwilldisplayinformationthatissummarized.

Adimensionalmodelwillbecreatedbyabusinessoperation.Departmentswithintheorganizationdonotplayaroleinthis.Thedatawillbepopulatedonceitisplacedwithinthedimensionalwarehouse,andisnotdependentonthevariousdepartmentsthatmaycomposeanorganization.Whenbusinessprocesseshavebeendevelopedwithinthewarehouse,thesystemwillbecomehighlyefficient.ThenextpopulardatawarehouseapproachthatyouwillwanttobecomefamiliarwithistheCorporateInformationFactory.AnothernameforthistechniqueistheEDWapproach.Thedatathatisextractedfromthesourcewillbecoordinated.

WithintheCIF,astandarddatawarehouseisusedtoholddatarepositories,anditmayalsohavespecificdatawarehouseswhicharedesignedfordatamining.Thedatamartsmaybedesignedforspecificdepartments,andtheymayhavesummarydatawhichisintheformofadimensionalstructure.Theatomicdatamaybeobtainedfromthestandarddatawarehouse.Whiletherearesomesimilaritiesbetweenthesetotechniques,therearesomenotabledifferencesaswell.

Oneoftheprimarydifferencesbetweenthesetwotechniquesisthenormalizeddatafoundation.WiththeKimballapproach,thedatastructuresthatmustbeobtainedbeforethedimensionalpresentationwillbedependentonthesourcedataandtransformation.Inmostcases,theduplicatestorageofdataisnotrequiredinbothdimensionalandnormalizedfoundations.Manyofthepeoplewhochoosetouseanormalizeddatastructurebelievethatitisfasterthanthedimensionalstructure,buttheyoftenfailtotakeETLintoconsideration.

Anotherthingthatseparatesthetwodatawarehouseapproachesisthemanagementofatomicdata.WiththeCIF,atomicdatawillbestoredwithinanormalizeddatawarehouse.Incontrast,theKimballmethodstatesthattheatomicdatashouldbeplacedwithinadimensionalstructure.Whenthedataisplacedwithinadimensionalstructure,itcanbesummarizedinawidevarietyofdifferentways.

Itisimportanttomakesuretheinformationyouhaveisdetailedsothatuserswillbeabletoaskrelevantquestions.Whilemostuserswillnotplaceanemphasisonthedetailsofoneatomictransaction,theymaywantasummaryofalargenumberoftransactions.Itisimportantforthemtohavethedetailssothattheywillbeabletoanswerimportantquestions.Theapproachthatyouchooseshouldbetheonewhichbestservestheneedsofyourcompany.3DataWarehouseDesignStrategies

Tobuildaneffectivedatawarehouse,itisimportantforyoutounderstanddatawarehousedesignprinciples.Ifyourdatawarehouseisnotbuiltcorrectly,youcanrunintoanumberofdifferentproblems.

Thepropermethodsforbuildingapowerfuldatawarehousearebasedoninformationtechnologytactics.Firstoff,itisimportantthatyouandyourorganizationunderstandtheimportanceofhavingadatawarehouse.Ifworkersfeelthatadatawarehouseisunnecessary,theymaynotuseit,andthiscouldcauseconflicts.Everyoneinyourorganizationshouldunderstandtheimportanceofusingthesystem.

Afteryouhavegotyourcolleaguesbehindtheconceptofusingadatawarehouse,youwillwanttonextfocusondataintegrity.Youwillwanttoavoiddesigningadatawarehousethatwillloaddatathatisnotconsistent.Itisalsoimportanttoavoidcreatingadatabasethatwillreplicatedata.Thegoalofyourorganizationshouldbetointegratedataandcreatestandardsthatwillbeusedandfollowed.Afterdataintegrity,youwillnextwanttolookatimplementationefficiency.Thisbasicallymeansthatyouwillwanttodesignatsystemthatissimpletouse.Itdoesn’tmatterhowwelldesignedyourdatawarehouseisifyourworkershaveahardtimeusingit.

Ifyourworkershaveahardtimeusingthedatawarehouse,itwillslowdownthespeedandproductivityofyouroperation.Whenitcomestocreatingadatawarehouse,youwillwanttomakeitassimpleaspossible.Allofyourworkersshouldbeabletouseitwithoutproblems.Implementationefficiencyisaprinciplethatnaturallyleadstothenexttopicyouwillwanttofocuson,andthisisuserfriendliness.Thisisaconceptthatisanimportantpartofyourbusiness.Thereasonforthisisbecauseenduserswillnotutilizeaprogramthatistoodifficulttouse.Itisimportantforyoutokeeptheminmind.Useadesignwhichisfriendlyandeasytolearn.

Onceyouhavedesignedadatawarehousethatisuserfriendly,youwillnextwanttolookatoperationalefficiency.Oncethedatawarehousehasbeencreated,itshouldbeabletocarryoutoperationsquickly.Inadditiontothis,itshouldnothaveerrorsorothertechnicalproblems.Whenerrorsortechnicalproblemsdooccur,theyshouldbesimpletofix.Anotherthingyouwillwanttolookatisthecostinvolvedwithsupportingthesystem.Youwillwanttokeepthesecostslowasmuchaspossible.

Thedesignprinciplesthathavebeendiscussedinthisarticlesofararemorerelatedtobusinessthaninformationtechnology.However,thereareanumberofITdesignprinciplesthatyouwillwanttofollow.Oneoftheseisscalability.Thisisaproblemthatmanydatawarehousedesignersruninto.Thebestwaytodealwiththisissueistocreateadatawarehousethatisscalablefromthebeginning.Designitinawaywhichwillallowittosupportexpansionsorupgrades.Youshouldbeabletoadaptittoanumberofdifferentbusinesssituations.Thebestdatawarehousesarethosewhicharescalable.

Thedatawarehousethatyoudesignshouldfallundertheguidelinesofinformationtechnologystandards.EverytoolthatyouusetobuildyourdatawarehouseshouldworkwellwithITstandards.Youwillwanttomakesureitisdesignedinawaythatmakesiteasierforyourworkerstouse.Whilefollowingtheguidelinesinthisarticlewon’tallowyoutoalwaysbesuccessful,itwillgreatlytiptheoddsinyourfavor.Youshouldbewaryofcompaniesthatpromiseyouperfectresultsifyouusetheirdesignmethods.[2]Nomatterhowwelldesignedyourdatawarehouseis,youwillalwaysrunintoproblems.However,followingtherightprincipleswillmaketheproblemseasiertorecognizeandsolve.

Whenitcomestousingadatawarehouse,itisnotamatterof“if”youwillrunintoproblems.Itismatterof“how”and“when”.Whenyourdatawarehouseiswelldesigned,youwillbebetterequippedtosolveanyproblemsyouencounter.

1.?warehousen.倉(cāng)庫(kù),貨棧。

2.?goover受歡迎,獲得接受；檢查。

3.?orientvt.vi.使熟悉,使適應(yīng)；使朝向；確定位置；朝向；確定方向；使適應(yīng)n.東方,亞洲。

4.?variantn.變體;變種;變型adj.不同的;差別的;變異的;各種各樣的。

5.?specificadj.明確的,確切的,詳盡的；具體的,特有的,特定的；僅限于……的。Vocabulary

6.?volatileadj.飛行的,揮發(fā)性的,可變的,不穩(wěn)定的,輕快的,爆炸性的n.有翅的動(dòng)物,揮發(fā)物。

7.?scheman.概要,計(jì)劃,圖表，模式。

8.?acquisitionn.獲得,得到的東西；得到的人,買進(jìn)。

9.?aggregationn.集合,凝聚,集成,集結(jié)(作用),集合[成]體,集團(tuán)。

10.?strategyn.戰(zhàn)略(學(xué)),策略,計(jì)謀,作戰(zhàn)方針;智謀,手腕strategyandtactics戰(zhàn)略與戰(zhàn)術(shù)。

11.?Intricateadj.復(fù)雜的,錯(cuò)綜的,難以理解的。

12.?martn.市場(chǎng);貿(mào)易場(chǎng)所。

13.?repositoryn.倉(cāng)庫(kù)，儲(chǔ)藏所；儲(chǔ)物器皿，博物館；學(xué)識(shí)淵博的人；受人信賴的人，知己。

14.?Stagingn.舉行,進(jìn)行;配置，階變，級(jí),級(jí)組，分段運(yùn)輸;分級(jí)法。

15.?Populatevt.居住，使人口聚居于;移民于;殖民于人口稠密(稀少)的城市。

[1]Beingsubjectorientedmeansthatthedatawillprovideinformationaboutaspecificsubjectratherthantheinformationaboutthefunctionsofacompany.Becauseadatawarehouseissubjectoriented,itwillallowyoutoanalyzeinformationthatisconnectedtoaspecificsubject.Beingintegratedmeansthatthedatathatiscollectedwithinthedatawarehousecancomefromdifferentsources,butcanbecombinedintooneunitthatisrelevantandlogical.Havingatime-variantmeansthatalltheinformationwithinthedatawarehousecanbefoundwithagivenperiodoftime.ImportantSentences

所謂“面向主題”，就是數(shù)據(jù)將提供有關(guān)一個(gè)具體的主題的信息，而不是有關(guān)公司運(yùn)行的信息。由于數(shù)據(jù)倉(cāng)庫(kù)是面向主題的，因此它就允許你分析與具體主題相關(guān)的