版權說明:本文檔由用戶提供并上傳,收益歸屬內容提供方,若內容存在侵權,請進行舉報或認領
文檔簡介
BigDataWeipingChenTopicsWhatisBigData?Why‘BigData’isabigdeal?NoSQLvsSQLHowtoDealwithBigData?What’sHadoop/MapReduce?RDBMSvsHadoop/MapReduceBigdataplayers/SoftwareTools/PlatformsExamplesWhatIsBigData?CapturingandmanaginglotsofinformationWorkingwithmanynewtypesofdataStructure/UnstructuredExploitingthesemassesofinformationandnewdatatypeswithnewstylesofapplicationsBiggerthanTerabytesvolume,variety,velocity,variabilityWhy‘BigData’isabigDealBigdatadiffersfromtraditionalinformationinmind-bendingways:
NotknowingwhybutonlywhatThechallengewithleadershipisthatit’sverydrivenbygutinstinctinmostcasesAirtravelerscannowfigureoutwhichflightsarelikeliesttobeontime,thankstodatascientistswhotrackedadecadeofflighthistorycorrelatedwithweatherpatternsPublishersusedatafromtextanalysisandsocialnetworkstogivereaderspersonalizednews.healthcareisoneofthebiggestopportunities,IfwehadelectronicrecordsofAmericansgoingbackgenerations,we'dknowmoreaboutgeneticpropensities,correlationsamongsymptoms,andhowtoindividualizetreatments.Googlemapsearchcorrelateto“Openretailstoreetc.”WhatThisMeansforYou
BigDatacanhelpacompanydomanythings:ProfilecustomersDeterminepricingstrategiesIdentifycompetitiveadvantagesBettertargetadvertisingInforminternalresearchandproductdevelopmentStrengthencustomerserviceMainstepsinadoptingananalyticalsystemWhatWillWeAnalyze?DoWeBuyorBuild?AreWeReadytoInvest?DoWeUnderstandtheImpact?ChallengesInformationgrowthProcessingpowerPhysicalstoragediskcapacityincreasedramatically100MB/Sreadfromdisk(bottleneck)dataseekingtimeisslowthandatatransferringDataissuesCostsRecentlyITTrendCommodityhardwareDistributedfilesystemsOpensourceoperatingsystems,databases,andotherinfrastructureSignificantlycheaperstorageService-orientedarchitecture
BigDataChainCollectDataIngest/CleanData(OriginallyETL.Existingschema)Humanexploration/Infrastructure/DataminingStore/ArchiveShare(decisionmake,othersystem)Measure/feedbackACIDACID(Atomicity,Consistency,Isolation,Durability)
(A)whenyoudosomethingtochangeadatabasethechangeshouldworkorfailasawhole(C)thedatabaseshouldremainconsistent(thisisaprettybroadtopic)(I)ifotherthingsaregoingonatthesametimetheyshouldn'tbeabletoseethingsmid-update(D)ifthesystemblowsup(hardwareorsoftware)thedatabaseneedstobeabletopickitselfbackup;andifitsaysitfinishedapplyinganupdate,itneedstobecertainMapReduceDividingandconqueringHighlyfaulttolerantnodesareexpectedtofail?Everydatablock(bydefault)replicatedon3nodes(isalsorackaware)DifficulttoimplementRDBMSfixed-schema,row-orienteddatabaseswithACIDpropertiesandasophisticatedSQLqueryengine.Theemphasisisonstrongconsistency,referentialintegrity,abstractionfromthephysicallayer,andcomplexqueriesthroughtheSQLlanguage.easilycreatesecondaryindexes,performcomplexinnerandouterjoins,count,sum,sort,group,andpageyourdataacrossanumberoftables,rows,andcolumns.RDBMSvsMapReduceRDBMSMapReducemostlystructureddataunstructureddatadatainternalstructurenone(doesinprocess)normalizedneednon-nomalizeNotes:1.relationaldatabasesstartincorporatingsomeoftheideasfromMapReduce(suchasAsterData’sandGreenplum’sdatabases)2.theotherdirection,ashigher-levelquerylanguagesbuiltonMapReduce(suchasPigandHive)makeMapReducesystemsmoreapproachablefortraditionaldatabaseprogrammers.ArchitechuresHowdoesMapReduceworkHDFS(HadoopDistributedFileSystem)
DataisstoredonlocaldiskandprocessingisdonelocallyonthecomputerwiththedataCanworkwithrawdatastoredinfilesystemordatabaseTwosteps:MapandReduce
MapMapReduceuseskey/valuepairs.(Traditionallyusingrowsandcolumns)
Example:lastname/chen
withdrawamount/20
transactiondate/06-23-2013Reducealltheintermediatevaluesforagivenoutputkeyarecombinedtogetherintoalist.Thereduce()functionthencombinestheintermediatevaluesintooneormorefinalvaluesforthesamekey.HadoopHadoopisdesignedtoabstractawaymuchofthecomplexityofdistributedprocessingDifferentfromGRIDcomputingWidelyusedSocialmedia(e.g.,Facebook,Twitter)
Lifesciences
Financialservices
Retail
GovernmentHadoopArchitectureApplicationlayer/enduseraccesslayera.JobTracker(workloadmanagementlayer)b.Distributedparallelfilesystems/datalayerHadoopImplementationHadoopisdesignedtorunjobsthatlastminutesorhoursontrusted,dedicatedhardwarerunninginasingledatacenterwithveryhighaggregatebandwidthinterconnectsDesignofHDFSNamenodes(TheMaster)Managemetadata/filetreesDatanodes(Workers)
store/retrievedatablockDatanodesdonotuseRAIDdisk.HDFSround-robinsHDFSblocksbetweenalldisks.RAIDlimitedbytheslowestdiskonthearray.
LimitationsofHDFSLow-latencydataaccessLotsofsmallfilesMultiplewriters,arbitraryfilemodificationsHDFSBlock64MB/128MB(normaldiskblock512KB).minimize‘seek’timefixedsizeratherthanfile,easystorage/replication%hadoopfsck/-files–blocks%hadoopfs–help(regularfilesystemoperation)%hadoopfs-copyFromLocalinput/docs/quangle.txthdfs://localhost/user/tom/quangle.txt%hadoopfs-mkdirbooks%hadoopfs-lsDataflowsFormatandTypesMapReducemodelindetail,and,inparticular,howdatainvariousformats,fromsimpletexttostructuredbinaryobjects,canbeusedwiththismodelmap:(K1,V1)→list(K2,V2)reduce:(K2,list(V2))→list(K3,V3)TextfileOnthetopoftheCrumpettyTreeTheQuangleWanglesat,Buthisfaceyoucouldnotsee,OnaccountofhisBeaverHat.isdividedintoonesplitoffourrecords.Therecordsareinterpretedasthefollowingkey-valuepairs:(0,OnthetopoftheCrumpettyTree)
(33,TheQuangleWanglesat,)(57,Buthisfaceyoucouldnotsee,)(89,OnaccountofhisBeaverHat.)DataFileMapreduceSpecialFeatureCounterSortingJoinsShuffle
MapReduceguaranteesthattheinputtoeveryreducerissortedbykey.Theprocessbywhichthesystemperformsthesort—andtransfersthemapoutputstothereducersasinputs-ShuffleInstallHadoop%cd/usr/local%sudotarxzfhadoop-x.y.z.tar.gzchangetheowneroftheHadoopfilestobethehadoopuserandgroup:%sudochown-Rhadoop:hadoophadoop-x.y.zLayers/Players--continueExtract,transform,load(ETL)
IBMInfoSphereDataStageInformaticaPervasiveTalendDatawarehouse
Oracle,Teradata,IBMNetezza,Greenplum
PIG–HelpHadoopPigisascriptinglanguageforexploringlargedatasetsAPigLatinprogramismadeupofaseriesofoperations,ortransformations,thatareappliedtotheinputdatatoproduceoutput2.PigexecutionenvironmenttranslatesintoanexecutablerepresentationandthenrunsHbaseHBaseisadistributedcolumn(family)-orienteddatabasebuiltontopofHDFS.HBaseistheHadoopapplicationtousewhenyourequirereal-timeread/writerandom-accesstoverylargedatasetsHBasetablesarelikethoseinanRDBMS,onlycellsareversioned,rowsaresorted,andcolumnscanbeaddedontheflybytheclientaslongasthecolumnfamilytheybelongtopreexists.Hbase--continueRegions
Eachregioncomprisesasubsetofatable’srowsprovidewaystoreadorwriteindividualrecordsefficientlybasedonHadoopHiveHive—anopensourcedatawarehousingandSQLinfrastructurebuiltontopofHadoopCloudera’sDistributionforHadoopCloudera’sDistributionforHadoopisbasedonthemostrecentstableversionofApacheHadoopwithnumerouspatches,backports,andupdatesEvaluateCriteriaHighscalabilityLowlatencyPredictabilityHighavailabilityEasymanagementMulti-tenancyBigDataRealtimeProcessingGoogleBigQueryisawebservicethatletsyoudointeractiveanalysisofmassivedatasets—uptobillionsofrowsTwitter’sStormClouderaImpalaNoSQLNoSQLreferstodocument-orienteddatabasesSQLdoesn’tscalewellhorizontally(addmoreserverswhichCloudisgoodat)Itisschemaless.Butnotformless(JSONformat).JSON:datainterchangeformatMongoDatabaseCouchDatabaseNoSQLBaseModelBaseModelBasicAvailability:spreaddataacrossmanystoragesystemswithahighdegreeofreplicationSoftState:dataconsistencyisthedeveloper'sproblemandshouldnotbehandledbythedatabase.EventualConsistency:atsomepointinthefuture,datawillconvergetoaconsistentstate.Noguaranteesaremade“when”JSONStructure{field1:value1,field2:value2…fieldN:valueN}varmydoc={_id:ObjectId("5099803df3f4948bd2f98391"),name:{first:"Alan",last:"Turing"},birth:newDate('Jun23,1912'),death:newDate('Jun07,1954'),contribs:["Turingmachine","Turingtest",…],views:NumberLong(1250000)}RDBMSvsNoSQLXszcRowDB:001:10,Smith,Joe,40000;002:12,Jones,Mary,50000;003:11,Johnson,Cathy,44000;004:22,Jones,Bob,55000;index:001:40000;002:50000;003:44000;004:55000;ColumnDB:10:001,12:002,11:003,22:004;Smith:001,Jones:002,Johnson:003,Jones:004;Joe:001,Mary:002,Cathy:003,Bob:004;40000:001,50000…;Smith:001,Jones:002,004,Johnson:003;…BenefitsColumn-orientedorganizationsaremoreefficientwhenanaggregateneedstobecomputedovermanyrowsbutonlyforanotablysmallersubsetofallcolumnsofdata,becausereadingthatsmallersubsetofdatacanbefasterthanreadingalldata.Column-orientedorganizationsaremoreefficientwhennewvaluesofacolumnaresuppliedforallrowsatonce,becausethatcolumndatacanbewrittenefficientlyandreplaceoldcolumndatawithouttouchinganyothercolumnsfortherows.Row-orientedorganizationsaremoreefficientwhenmanycolumnsofasinglerowarerequiredatthesametime,andwhenrow-sizeisrelativelysmall,astheentirerowcanberetrievedwithasinglediskseek.Row-orientedorganizationsaremoreefficientwhenwritinganewrowifallofthecolumndataissuppliedatthesametime,astheentirerowcanbewrittenwithasinglediskseek.SQLvsNonSQLAgoodcompromiseistodesignyoursystemwith3logicalDBs1.NormalSQLDBusedbyyouradminapplicationtocreatecontent.
2.No-SQLDBforfront-end/public/high-volumeapplicaitonusedbythepublicinternet.
3.ThelastDBisforanalyticalreportingsystemusingcubesandallthatgoodstuff.ThendataflowsfromtheAdminDBtotheclientNo-SQLDBwhensomeone"Publishes"apieceofcontent,theclient(NoSQL)dbprovidesveryfastreadaccessandrecordsuserinteractionswiththecontent.ThenyouhaveascheduledjobthatpullsthedatafromtheclientDBintothereportingsystem.SinceAdmin,client,andreportingareoftenseparateapps,eachapplicationteamcanworkwithdataintheformatthatbestservestheapplicationandthetransitionfromonesystemtotheotherishandledintheservicelayers.BigDataSolutionsCloudera:ClouderaEnterpriseMicrosoft:WindowsAzureHDInsightServiceGoogle:BigQueryAmazon:DynamoDBIBM:InfoSphereStreams/NetezzaEMC:GreenplumTeraData:AsterMapReducePlatformOracle:Hadoop/MapreduceBigDataconnectorsBigDataProjectFailReasonsLackofcooperationamongdepartmentsLackof
staff
experiencedinBigDataSecurityPoorplanningRealExamplesofBigDataProjectsConsumerproductcompaniesandretailorganizationsaremonitoringsocialmedialikeFacebookandTwittertogetanunprecedentedviewintocustomerbehavior,preferences,andproductperception.Manufacturersaremonitoringminutevibrationdatafromtheirequipment,whichchangesslightlyasitwearsdown,topredicttheoptimaltimetoreplaceormaintain.Replacingittoosoonwastesmoney;replacingittoolatetriggersanexpensiveworkstoppageManufacturersarealsomonitoringsocialnetworks,butwithadifferentgoalthanmarketers:Theyareusingittodetectaftermarketsupportissuesbeforeawarrantyfailurebecomespubliclydetrimental.FinancialServicesorganizationsareusingdataminedfromcustomerinteractionstosliceanddicetheirusersintofinelytunedsegments.Thisenablesthesefinancialinstitutionstocreateincreasinglyrelevantandsophisticatedoffers.ContinuationAdvertisingandmarketingagenciesaretrackingsocialmediatounderstandresponsivenesstocampaigns,promotions,andotheradvertisingmediums.InsurancecompaniesareusingBigDataanalysistoseewhichhomeinsuranceapplicationscanbeimmediatelyprocessed,andwhichonesneedavalidatingin-personvisitfromanagent.Byembracingsocialmedia,retailorganizationsareengagingbrandadvocates,changingtheperceptionofbrandantagonists,andevenenablingenthusiasticcustomerstoselltheirproducts.Hospitalsareanalyzingmedicaldataandpatientrecordstopredictthosepatientsthatarelikelytoseekreadmissionwithinafewmonthsofdischarge.Thehospitalcantheninterveneinhopesofpreventinganothercostlyhospitalstay.Web-basedbusinessesaredevelopinginformationproductsthatcombinedatagatheredfromcustomerstooffermoreappealingrecommendationsandmoresuccessfulcouponprograms.Thegovernmentismakingdatapublicatboththenational,state,andcitylevelforuserstodevelopnewapplicationsthatcangeneratepublicgood.Sportsteamsareusingdatafortrackingticketsalesandevenfortrackingteamstrategies.StartingBigDataProjectsNYTD(NationalYouthinTransitionDatabase)DocumentationSearchDynamicSQLtableWWWlogfilesHealthCare:extractingnames,locations,dates,products,diseases,Rx,conditions,etc.,fromtextNYTD(NationalYouthTransitinalDatabase)DatacolectionsystemtotracktheStatesaretocollectinformationoneachyouthwhoreceivesindependentlivingservicespaidfororprovidedbytheStateagencythatadministerstheCFCIP.Second,StatesaretocollectdemographicandoutcomeinformationoncertainyouthinfostercarewhomtheStatewillfollowovertimetocollectadditionaloutcomeinformationthe
溫馨提示
- 1. 本站所有資源如無特殊說明,都需要本地電腦安裝OFFICE2007和PDF閱讀器。圖紙軟件為CAD,CAXA,PROE,UG,SolidWorks等.壓縮文件請下載最新的WinRAR軟件解壓。
- 2. 本站的文檔不包含任何第三方提供的附件圖紙等,如果需要附件,請聯系上傳者。文件的所有權益歸上傳用戶所有。
- 3. 本站RAR壓縮包中若帶圖紙,網頁內容里面會有圖紙預覽,若沒有圖紙預覽就沒有圖紙。
- 4. 未經權益所有人同意不得將文件中的內容挪作商業(yè)或盈利用途。
- 5. 人人文庫網僅提供信息存儲空間,僅對用戶上傳內容的表現方式做保護處理,對用戶上傳分享的文檔內容本身不做任何修改或編輯,并不能對任何下載內容負責。
- 6. 下載文件中如有侵權或不適當內容,請與我們聯系,我們立即糾正。
- 7. 本站不保證下載資源的準確性、安全性和完整性, 同時也不承擔用戶因使用這些下載資源對自己和他人造成任何形式的傷害或損失。
最新文檔
- 2024年度某物流公司與某制造商關于物流服務合同3篇
- 2024年暑期實習生實習合同及人才儲備協議3篇
- 2024年度地板打蠟與綠色環(huán)保產業(yè)合作合同3篇
- 2024年度防盜門安裝與社區(qū)安全保障體系合同2篇
- 2024年度土地承包合同終止后的土地復耕補貼協議3篇
- 2024年房屋買賣包含土地使用權合同3篇
- 2024年大蒜種植基地與農產品加工企業(yè)采購供應合同3篇
- 2024年度廣告發(fā)布合同:某品牌廣告在社交媒體平臺發(fā)布3篇
- 2024版保險代理協議合同3篇
- 2024版企業(yè)法律視角下的人力資源績效評估合同3篇
- 空白貨品簽收單
- 水泥混凝土路面施工方案85171
- 建筑電氣施工圖(1)課件
- 質量管理體系運行獎懲考核辦法課案
- 泰康人壽養(yǎng)老社區(qū)介紹課件
- T∕CSTM 00584-2022 建筑用晶體硅光伏屋面瓦
- 2020春國家開放大學《應用寫作》形考任務1-6參考答案
- 國家開放大學實驗學院生活中的法律第二單元測驗答案
- CAMDS操作方法及使用技巧
- Zarit照顧者負擔量表
- 2021年全國質量獎現場匯報材料-財務資源、財務管理過程及結果課件
評論
0/150
提交評論