3)MapReduce Simplied Data Processing on Large Clusters1.PDF

3)MapReduce Simplied Data Processing on Large Clusters1.PDF

ID:34815676

大小:194.82 KB

页数:13页

时间:2019-03-11

3)MapReduce Simplied Data Processing on Large Clusters1.PDF_第1页
3)MapReduce Simplied Data Processing on Large Clusters1.PDF_第2页
3)MapReduce Simplied Data Processing on Large Clusters1.PDF_第3页
3)MapReduce Simplied Data Processing on Large Clusters1.PDF_第4页
3)MapReduce Simplied Data Processing on Large Clusters1.PDF_第5页
资源描述:

《3)MapReduce Simplied Data Processing on Large Clusters1.PDF》由会员上传分享,免费在线阅读,更多相关内容在学术论文-天天文库

1、MapReduce:SimpliedDataProcessingonLargeClustersJeffreyDeanandSanjayGhemawatjeff@google.com,sanjay@google.comGoogle,Inc.Abstractgivenday,etc.Mostsuchcomputationsareconceptu-allystraightforward.However,theinputdataisusuallyMapReduceisaprogrammingmodelandanassoci-largeandthec

2、omputationshavetobedistributedacrossatedimplementationforprocessingandgeneratinglargehundredsorthousandsofmachinesinordertonishindatasets.Usersspecifyamapfunctionthatprocessesaareasonableamountoftime.Theissuesofhowtopar-key/valuepairtogenerateasetofintermediatekey/valueall

3、elizethecomputation,distributethedata,andhandlepairs,andareducefunctionthatmergesallintermediatefailuresconspiretoobscuretheoriginalsimplecompu-valuesassociatedwiththesameintermediatekey.Manytationwithlargeamountsofcomplexcodetodealwithrealworldtasksareexpressibleinthismode

4、l,asshowntheseissues.inthepaper.Asareactiontothiscomplexity,wedesignedanewProgramswritteninthisfunctionalstyleareautomati-abstractionthatallowsustoexpressthesimplecomputa-callyparallelizedandexecutedonalargeclusterofcom-tionsweweretryingtoperformbuthidesthemessyde-moditymac

5、hines.Therun-timesystemtakescareofthetailsofparallelization,fault-tolerance,datadistributiondetailsofpartitioningtheinputdata,schedulingthepro-andloadbalancinginalibrary.Ourabstractionisin-gram'sexecutionacrossasetofmachines,handlingma-spiredbythemapandreduceprimitivesprese

6、ntinLispchinefailures,andmanagingtherequiredinter-machineandmanyotherfunctionallanguages.Werealizedthatcommunication.Thisallowsprogrammerswithoutanymostofourcomputationsinvolvedapplyingamapop-experiencewithparallelanddistributedsystemstoeas-erationtoeachlogical“record”inour

7、inputinordertoilyutilizetheresourcesofalargedistributedsystem.computeasetofintermediatekey/valuepairs,andthenOurimplementationofMapReducerunsonalargeapplyingareduceoperationtoallthevaluesthatsharedclusterofcommoditymachinesandishighlyscalable:thesamekey,inordertocombinethed

8、eriveddataap-atypicalMapReducecomputationprocessesmanyter-propriately.Ouruseofafun

当前文档最多预览五页,下载文档查看全文

此文档下载收益归作者所有

当前文档最多预览五页,下载文档查看全文
温馨提示:
1. 部分包含数学公式或PPT动画的文件,查看预览时可能会显示错乱或异常,文件下载后无此问题,请放心下载。
2. 本文档由用户上传,版权归属用户,天天文库负责整理代发布。如果您对本文档版权有争议请及时联系客服。
3. 下载前请仔细阅读文档内容,确认文档内容符合您的需求后进行下载,若出现内容与标题不符可向本站投诉处理。
4. 下载文档时可能由于网络波动等原因无法下载或下载错误,付费完成后未能成功下载的用户请联系客服处理。