外文翻译 - 数据库系统

0外文原文DataBaseSystem1、IntroductionThedevelopmentofcorporatedatabaseswillbeoneofthemostimportantdata-processingactivitiesfortherestofthe1970s.Datewillbeincreasinglyregardedasavitalcorporateresource,whichmustbeorganizedsoastomaximizetheirvalue.Inadditiontothedatabaseswithinanorganization,avastnewdemandisgrowingfordatabaseservices,whichwillcollect,organize,andselldata.Thefilesofdatawhichcomputerscanusearegrowingatastaggeringrate.Thegrowthrateinthesizeofcomputerstorageisgreaterthanthegrowthinthesizeorpowerofanyothercomponentintheexplodingdataprocessingindustry.Themoredatathecomputershaveaccessto,thegreateristheirpotentialpower.Inallwalksoflifeandinallareasofindustry,databankswillchangetheareasofwhatitispossibleformantodo.Intheendofthiscentury,historianswilllookbacktothecomingofcomputerdatabanksandtheirassociatedfacilitiesasastepwhichchangedthenatureoftheevolutionofsociety,perhapseventuallyhavingagreatereffectonthehumanconditionthaneventheinventionoftheprintingpress.Somemostimpressivecorporategrowthstoriesofthegenerationarelargelyattributabletotheexplosivegrowthintheneedofinformation.Thevastmajorityofthisinformationisnotyetcomputerized.However,thecostofdatastoragehardwareisdroppingmorerapidlythanothercostsindataprocessing.Itwillbecomecheapertostoredataoncomputerfilesthantostorethemonpaper.Notonlyprintedinformationwillbestored.Thecomputerindustryisimprovingitscapabilitytostorelinedrawing,datainfacsimileform,photo-graphs,humanspeech,etc.Infact,anyformofinformationotherthanthemostintimatecommunicationsbetweenhumanscanbetransmittedandstoreddigitally.Therearetwomaintechnologydevelopmentslikelytobecomeavailableinthenearfuture.First,thereareelectromagneticdevicesthatwillholdmuchmoredatathandisksbut1havemuchlongeraccesstime.Second,therearesolid-statetechnologiesthatwillgivemicrosecondaccesstimebutcapacitiesaresmallerthandisks.Disksthemselvesmaybeincreasedincapacitysomewhat.Forthelongertermfuturethereareanumberofnewtechnologieswhicharecurrentlyworkinginresearchlabswhichmayreplacedisksandmayprovideverylargemicrosecond-access-timedevices.Asteadystreamofnewstoragedevicesisthuslikelytoreachthemarketplaceoverthenext5years,rapidlyloweringthecostofstoringdata.Giventheavailabletechnologies,itislikelythaton-linedatabaseswillusetwoorthreelevelsofstorage.Onesolid-statewithmicrosecondaccesstime,oneelectromagneticwithaccesstimeofafractionofasecond.Iftwo,three,orfourlevelsofstorageareused,physicalstorageorganizationwillbecomemorecomplex,probablywithpagingmechanismstomovedatabetweenthelevels;solid-statestorageoffersthepossibilityofparallelsearchoperationandassociativememory.Boththequantityofdatastoredandthecomplexityoftheirorganizationaregoingupbyleapsandbounds.Thefirsttrillionbiton-linestoresarenowinuse.inafewyearstime,storesofthissizemaybecommon.Aparticularlyimportantconsiderationindatabasedesignistostorethedatasothatthecanbeusedforawidevarietyofapplicationsandsothatthewaytheycanbechangedquicklyandeasily.Oncomputerinstallationpriortothedatabaseeraithasbeenremarkablydifficulttochangethewaydataareused.Differentprogrammersviewthedataindifferentwaysandconstantlywanttomodifythemasnewneedsarisemodification,however,cansetoffachainreactionofchangestoexistingprogramsandhencecanbeexceedinglyexpensivetoaccomplish.Consequently,dataprocessinghastendedtobecomefrozenintoitsolddatastructures.Toachieveflexibilityofdatausagethatisessentialinmostcommercialsituations.Twoaspectsofdatabasedesignareimportant.First,itshouldbepossibletointerrogateandsearchthedatabasewithoutthelengthyoperationofwritingprogramsinconventionalprogramminglanguages.Second,thedatashouldbeindependentoftheprogramswhich2usethemsothattheycanbeaddedtoorrestructuredwithouttheprogramsbeingchanged.Theworkofdesigningadatabaseisbecomingincreasingdifficult,especiallyifitistoperforminanoptimalfashion.Therearemanydifferentwaysinwhichdatacanbestructured,andtheyhavedifferenttypesofdataneedtobeorganizedindifferentways.Differentdatahavedifferentcharacteristics,whichoughttoeffectthedataorganization,anddifferentusershavefundamentallydifferentrequirements.Soweneedakindofdatabasemanagementsystem(DBMS)tomanagedata.Databasedesignusingtheentity-relationshipmodelbeginswithalistoftheentitytypesinvolvedandtherelationshipsamongthem.Thephilosophyofassumingthatthedesignerknowswhattheentitytypesareattheoutsetissignificantlydifferentfromthephilosophybehindthenormalization-basedapproach.Theentity-relationship(E-R)approachusesentity-relationshipdiagrams.TheE-RapproachrequiresseveralstepstoprodureastructurethatisacceptablebytheparticularDBMS.Thesestepsare:(1)Dataanalysis(2)Producingandoptimizingtheentitymodel.(3)Logicalschemadevelopment(4)Physicaldatabasedesignprocess.Developingadatabasestructurefromuserrequirementsiscalleddatabasesdesign.Mostpractitionersagreethattherearetwoseparatephasestothedatabasedesignprocess.Thedesignofalogicaldatabasestructurethatisprocessablebythedatabasemanagementsystem(DBMS)describestheusersviewofdata,andistheselectionofaphysicalstructuresuchastheindexedsequentialordirectaccessmethodoftheintendedDBMS.Currentdatabasedesigntechnologyshowsmanyresidualeffectsofitsoutgrowthfromsingle-recordfiledesignmethods.Filedesignisprimarilyapplicationprogramdependentsincethedatahasbeendefinedandstructuredintermsofindividualapplicationstousethem.TheadventofDBMSrevisedtheemphasisindataandprogramdesignapproaches.Therearemanyinterlockingquestionsinthedesignofdata-basesystemsandmany3typesoftechniquethatonecanuseisanswertothequestionsomany;infact,thatoneoftenseesvaluableapproachesbeingoverlookedinthedesignandvitalquestionsnotbeingasked.Therewillsoonbenewstoragedevices,newsoftwaretechniques,andnewtypesofdatabases.Thedetailswillchange,butmostoftheprincipleswillremain.Therefore,thereadershouldconcentrateontheprinciples.2、DatabasesystemTheconceptionusedfordescribingfilesanddatabaseshasvariedsubstantiallyinthesameorganization.Adatabasemaybedefinedasacollectionofinterrelateddatastoredtogetherwithaslittleredundancyaspossibletoserveonormoreapplicationsinanoptimalfashion;thedataarestoredsothattheyareindependentofprogramswhichusethedata;acommonandcontrolledapproachisusedinaddingnewdataandinmodifyingandretrievingexistingdatawithinthedatabase.Onesystemissaidtocontainacollectionofdatabasesiftheyareentirelyseparateinstructure.Adatabasemaybedesignedforbatchprocessing,real-timeprocessing,orin-lineprocessing.Adatabasesysteminvolveapplicationprogram,DBMS,anddatabase.Oneofthemostimportantcharacteristicsofmostdatabasesisthattheywillconstantlyneedtochangeandgrow.Easyrestructuringofthedatabasemustbepossibleasnewdatatypesandnewapplicationsareadded.Therestructuringshouldbepossiblewithouthavingtorewritetheapplicationprogramandingeneralshouldcauseaslittleupheavalaspossible.Theeasewithwhichadatabasecanbechangedwillhaveamajoreffectontherateatwhichdata-processingapplicationcanbedevelopedinacorporation.Thetermdataindependenceisoftenquotedasbeingoneofthemainattributesofadatabase.Itimpliesthatthedataandtheapplicationprogramswhichusethemareindependentsothateithermaybechangedwithoutchangingtheother.Whenasinglesetofdataitemsservesavarietyofapplications,differentapplicationprogramsperceivedifferentrelationshipsbetweenthedataitems.Toalargeextent,data-baseorganizationisconcernedwiththerepresentationofrelationshipbetweendataitemsandrecordsaswell4ashowandwherethedataarestored.Adatabaseusedformanyapplicationscanhavemultipleinterconnectionsbetweenthedataitemaboutwhichwemaywishtorecord.Itcandescribestherealworld.Thedataitemrepresentsanattribute,andtheattributemustbeassociatedwiththerelevantentity.Wedesignvaluestotheattributes,oneattributehasaspecialsignificanceinthatitidentifiestheentity.Anattributeorsetofattributewhichthecomputerusestoidentifyarecordortupleisreferredtoasakey.Theprimarykeyisdefinedasthatkeyusedtouniquelyidentifyonerecordortuple.Theprimarykeyisofgreatimportancebecauseitisusedbythecomputerinlocatingtherecordortuplebymeansofanindexoraddressingalgorithm.Ifthefunctionofadatabaseweremerelytostoredata,itsorganizationwouldbesimple.Mostofthecomplexitiesarisefromthefactthatismustalsoshowtherelationshipsbetweenthevariousitemsofdatathatarestored.Itisdifferenttodescribethedatainlogicalorphysical.Thelogicaldatabasedescriptionisreferredtoasaschema.Aschemaisachartofthetypesofdatathatoneused.Itgivesthenamesoftheentitiesandattributes,andspecificstherelationsbetweenthem.Itisaframeworkintowhichthevaluesofthedata-itemscanbefitted.Wemustdistinguishbetweenarecordtypeandainstanceoftherecord.Whenwetalkabouta“personnelrecord”,thisisreallyarecordtype.Therearenodatavaluesassociatedwithit.Thetermschemaisusedtomeananoverallchartofallofthedataitemtypesandrecordtypesstoredinadataheuses.Manydifferentsubschemacanbederivedfromoneschema.Theschemaandthesubschemaarebothusedbythedata-basemanagementsystem,theprimaryfunctionofwhichistoservetheapplicationprogramsbyexecutingtheirdataoperations.ADBMSwillusuallybehandingmultipledatacallsconcurrently.Itmustorganizeitssystembufferssothatdifferentdataoperationscanbeinprocesstogether.Itprovidesadatadefinitionlanguagetospecifytheconceptualschemaandmostlikely,someofthe5detailsregardingtheimplementationoftheconceptualschemabythephysicalschema.Thedatadefinitionlanguageisahigh-levellanguage,enablingonetodescribetheconceptualschemaintermsofa“datamodel”.Thechoiceofadatamodelisadifficultone,sinceitmustberichenoughinstructuretodescribesignificantaspectsoftherealworld,yetitmustbepossibletodeterminefairlyautomaticallyanefficientimplementationoftheconceptualschemabyaphysicalschema.ItshouldbeemphasizedthatwhileaDBMSmightbeusedtobuildsmalldatabases,manydatabasesinvolvemillionsofbytes,andaninefficientimplementationcanbedisastrous.Wewilldiscussthedatamodelinthefollowing.3、ThreeDataModelsLogicalschemasaredefinedasdatamodelswiththeunderlyingstructureofparticulardatabasemanagementsystemssuperimposedonthem.Atthepresenttime,therearethreemainunderlyingstructuresfordatabasemanagementsystems.Theseare:RelationalHierarchicalNetworkThehierarchicalandnetworkstructureshavebeenusedforDBMSsincethe1960s.Therelationalstructurewasintroducedintheearly1970s.Intherelationalmodel,theentitiesandtheirrelationshipsarerepresentedbytwo-dimensionaltables.Everytablerepresentsanentityandismadeupofrowsandcolumns.Relationshipsbetweenentitiesarerepresentedbycommoncolumnscontainingidenticalvaluesfromadomainorrangeofpossiblevalues.Thelastuserispresentedwithasimpledatamodel.Hisandherrequestareformulatedintermsoftheinformationcontentanddonotreflectanycomplexitiesduetosystem-orientedaspects.Arelationaldatamodeliswhattheusersees,butitisnotnecessarilywhatwillbeimplementedphysically.Therelationaldatamodelremovesthedetailsofstoragestructureandaccessstrategyfromtheuserinterface.Themodelprovidesarelativelyhigherdegreeofdata.Tobeabletomakeuseofthispropertyoftherelationaldatamodelhowever,thedesignoftherelations6mustbecompleteandaccurate.AlthoughsomeDBMSbasedontherelationaldatamodelarecommerciallyavailabletoday,itisdifficulttoprovideacompletesetofoperationalcapabilitieswithrequiredefficiencyonalargescale.Itappearstodaythattechnologicalimprovementsinprovidingfasterandmorereliablehardwaremayanswerthequestionpositively.Thehierarchicaldatamodelisbasedonatree-likestructuremadeupofnodesandbranches.Anodeisacollectionofdataattributesdescribingtheentityatthatpoint.Thehighestnodeofthehierarchicaltreestructureiscalledaroot.Thenodesatsucceedinglowerlevelsarecalledchildren.Ahierarchicaldatamodelalwaysstartswitharootnode.Everynodeconsistsofoneormoreattributesdescribingtheentityatthatnode.Dependentnodescanfollowthesucceedinglevels.Thenodeintheprecedinglevelbecomestheparentnodeofthenewdependentnodes.Aparentnodecanhaveonechildnodeasadependentormanychildrennodes.Themajoradvantageofthehierarchicaldatamodelistheexistenceofprovendatabasemanagementsystemsthatusethehierarchicaldatamodelasthebasicstructure.Thereisareductionofdatadependencybutanychildnodeisaccessibleonlythroughitsparentnode,themany-tomanyrelationshipcanbeimplementedonlyinaclumsyway.Thisoftenresultsinaredundancyinstoreddata.Thenetworkdatamodelinterconnectstheentitiesofanenterpriseintoanetwork.Inthenetworkdatamodeladatabaseconsistsofanumberofareas.Anareacontainsrecords.Inturn,arecordmayconsistoffields.Asetwhichisagroupingofrecords,mayresideinanareaorspananumberofareas.Asettypeisbasedontheownerrecordtypeandthememberrecordtype.Themany-tomanyrelation-ship,whichoccursquitefrequentlyinreallifecanbeimplementedeasily.Thenetworkdatamodelisverycomplex,theapplicationprogrammermustbefamiliarwiththelogicalstructureofthedatabase.4、LogicalDesignandPhysicalDesignLogicaldesignofdatabasesismainlyconcernedwithsuperimposingtheconstructsofthedatabasemanagementsystemonthelogicaldatamodel.Therearethreemainlymodels:hierarchical,relational,networkwehavementionedabove.7Thephysicalmodelisaframeworkofthedatabasetobestoredonphysicaldevices.Themodelmustbeconstructedwitheveryregardgiventotheperformanceoftheresultingdatabase.Oneshouldcarryoutananalysisofthephysicalmodelwithaveragefrequenciesofoccurrencesofthegroupingsofthedataelements,withexpectedspaceestimates,andwithrespecttotimeestimatesforretrievingandmaintainingthedata.Thedatabasedesignermayfinditnecessarytohavemultipleentrypointsintoadatabase,ortoaccessaparticularsegmenttypewithmorethanonekey.Toprovidethistypeofaccess;itmaybenecessarytoinvertthesegmentonthekeys.ThephysicaldesignermusthaveexpertiseinknowledgeoftheDBMSfunctionsandunderstandingofthecharacteristicsofdirectaccessdevicesandknowledgeoftheapplications.Manydatabaseshavelinksbetweenonerecordandanother,calledpointers.Apointerisafieldinonerecordwhichindicateswhereasecondrecordislocatedonthestoragedevices.Recordsthatexistonstoragedevicesisagivenphysicalsequence.Thissequencingmaybeemployedforsomepurpose.Themostcommonpuposeisthatrecordsareneededinagivensequencebycertaindata-processingoperationsandsotheyarestoredinthatsequences.Differentapplicationsmayneedrecordsindifferentsequences.Themostcommonmethodoforderingrecordsistohavetheminsequencebyakeythatkeywhichismostcommonlyusedforaddressingthem.Anindexisrequiredtofindanyrecordwithoutalengthysearchofthefile.Ifthedatarecordsarelaidoutsequentiallybykey,theindexforthatkeycanbemuchsmallerthantheyarenonsequential.Hashinghasbeenusedforaddressingrandom-accessstoragessincetheyfirstcameintoexistenceinthemid-1950s.Butnobodyhadthetemeritytousethewordhashinguntil1968.Manysystemsanalysishasavoidedtheuseofhashinginthesuspicionthatitiscomplicated.Infact,itissimpletouseandhastwoimportantadvantagesoverindexing.First,itfindsmostrecordswithonlyoneseekandsecond,insertionanddeletionscanbe8handledwithoutaddedcomplexity.Indexing,however,canbeusedwithafilewhichissequentialbyprimekeyandthisisanoverridingadvantage,forsomebatch-pro-cessingapplications.Manydata-basesystemsusechainstointerconnectrecordsalso.Achainreferstoagroupofrecordsscatterswithinthefilesandinterconnectedbyasequenceofpointers.Thesoftwarethatisusedtoretrivethechainedrecordswillmakethemappeartotheapplicationprogrammerasacontiguouslogicalfile.Theprimarydisadvantageofchainedrecordsisthatmanyreadoperationsareneededinordertofollowlengthychains.Sometimesthisdoesnotmatterbecausetherecordshavetobereadanyway.Inmostsearchoperations,however,thechainshavetobefollowedthroughrecordswhichwouldnototherwisetoread.Insomefileorganizationsthechainscanbecontainedwithinblockedphysicalrecordssothatexcessivereadsdonotoccur.Ringshavebeenusedinmanyfileorganizations.Theyareusedtoeliminateredundancy.Whenaringorachainisenteredatapointsomedistancefromitshead,itmaybedesirabletoobtaintheinformationattheheadquicklywithoutsteppingthroughalltheinterveninglinks.5、DataDescriptionLanguagesItisnecessaryforboththeprogrammersandthedataadministratortobeabletodescribetheirdataprecisely;theydosobymeansofdatadescriptionlanguages.Adatadescriptionlanguageisthemeansofdeclaringtodata-basemanagementsystemwhatdatastructureswillbeused.Adatadescriptionlanguagesgivingalogicaldatadescriptionshouldperformthefolloeingfunctions:Itshouldgiveauniquenametoeachdata-itemtype,filetype,databaseandotherdatasubdivision.Itshouldidentifythetypesofdatasubdivisionsuchasdataitemsegment,recordandbasefile.Itmaydefinethetypeofencodingtheprogramusesinthedataitems(binary,character,bitstring,etc.)9Itmaydefinethelengthofthedataitemsandtherangeofthevaluesthatadataitemcanassume.Itmayspecifythesequenceofrecordsinafileorthesequenceofgroupsofrecordinthedatabase.Itmayspecifymeansofcheckingforerrorsinthedata.Itmayspecifyprivacylocksforpreventingunauthorizedreadingormodificationofthedata.Thesemayoperateatthedata-item,segment,record,fileordata-baselevelandifnecessarymaybeextendedtothecontents(value)ofindividualdataitems.Theauthorizationmay,ontheotherhand,beseparatedefined.Itismoresubjecttochangethanthedatastructures,andchangesinauthorizationproceduresshouldnotforcechangesinapplicationprograms.Alogicaldatadescriptionshouldnotspecifyaddressing,indexing,orsearchingtechniquesorspecifytheplacementofdataonthestorageunits,becausethesetopicsareinthedomainofphysical,notlogicalorganization.Itmaygiveanindicationofhowthedatawillbeusedorofsearchingrequirement.Sothatthephysicaltechniquecanbeselectedoptimallybutsuchindicationsshouldnotbelogicallylimiting.MostDBMShavetheirownlanguagesfordefiningtheschemasthatareused.Inmostcasesthesedatadescriptionlanguagesaredifferenttootherprogrammerlanguage,becauseotherprogrammerdonothavethecapabilitytodefinetovarietyofrelationshipthatmayexitintheschemas.10中文翻译数据库系统1、介绍在20世纪70年代后期综合数据库的发展是最重要的数据处理活动之一。数据越来越被视为一种重要的综合资源，它必须被组织以扩大他们的价值。另外，在组织内的数据库，对数据库服务将有大量的新要求，就是收集、组织和出售数据。计算机使用数据文件正以令人难以置信的速度增长。计算机存贮量大小的增长速度比爆炸式的数据处理工业中的任何其它元件在规模或能量上的增长速度大得多。计算机存取的数据越多，它们的潜能就越大，在所有社会生活及工业范围中，数据库将改变人们可能做到的领域，在本世纪末，历史学家回过来将把计算机数据库和其有关的设备的出现，作为改变社会发展的自然规律的步骤，或许比起印刷业的发明来，会对人类环境产生更大的影响。某些使人印象深刻的成长历程很大程度上归因于所需信息的爆长。这些信息的绝大多灵数尚未计算机化，然而，储存数据的硬件的花费比数据处理的其它花费更迅速地下降。而且数据储存在计算机文件上比储存在纸上更便宜，储存不仅仅是打印信息。计算机工业正在改进储存绘图、传真表格中的数据、照片、人类语言等等的能力。事实上，除了人类之间私下交流的信息以外，任何信息形式均可以数字化地传输和储存。在不久的将来，两种主要的技术有希望变得实用，首先，有一种电磁设备将比磁盘存贮更多的数据，但存取时间长；第二，有一种固态技术将给出微秒级存取时间，但容量比磁盘小。磁盘本身在某些能力上可能提高，在未来的一段长时期内，将有大量的目前还在研究室中服务的新技术代替磁盘，并可能提供更快的微秒级存取时间的设备，新储存设备的发展趋势稳定，因此很有希望在下一个五年中上市，并迅速地降低储存数据的成本。若给出可用的技