A Benchmark for MapReduce Framework

A Benchmark for MapReduce Framework

ID:39777527

大小:394.99 KB

页数:8页

时间:2019-07-11

A Benchmark for MapReduce Framework_第1页
A Benchmark for MapReduce Framework_第2页
A Benchmark for MapReduce Framework_第3页
A Benchmark for MapReduce Framework_第4页
A Benchmark for MapReduce Framework_第5页
资源描述:

《A Benchmark for MapReduce Framework》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、200814thIEEEInternationalConferenceonParallelandDistributedSystemsMRBench:ABenchmarkforMap-ReduceFrameworkKiyoungKim,KyunghoJeon,HyuckHan,Shin-gyuKim,HyungsooJung,HeonY.YeomSchoolofComputerScienceandEngineeringSeoulNationalUniversitySeoul151-742,Korea{kykim,khjeon,hhyuck,sgkim,jhs,yeom}@dcsla

2、b.snu.ac.krAbstractdatadistribution,parallelprocessing,resultscollectionwithfaulttolerance.TherearesomesolutionssuchasMapReduceisGoogle’sprogrammingmodelforeasyMPIandGridforparallelcomputing,butitisnoteasydevelopmentofscalableparallelapplicationswhichpro-tosolveproblemswithbothefficiencyandconv

3、enience.cesshugequantityofdataonmanyclusters.DuetoMapReduce[7]isGoogle’sprogrammingmodelitsconveniencyandefficiency,MapReduceisusedinforscalableparalleldataprocessing,andvariousvariousapplications(e.g.,websearchservicesandon-implementations[7,11,1]ofMapReduceareusedtolineanalyticalprocessing.)H

4、owever,thereareonlyfewsolveproblemsinmanyareas(i.e.,websearchser-goodbenchmarkstoevaluateMapReduceimplementa-vice.)MapReduceusestwofunctions-MapandRe-tionsbyrealistictestsets.duce-fromtheparallelizablefunctions.Usersim-Inthispaper,wepresentMRBenchthatisabench-plementonlyMapfunctionandReducefu

5、nctiontomarkforevaluatingMapReducesystems.MRBenchfo-converttheirprogramstoparallelizedversionseasily.cusesonprocessingbusinessorientedqueriesandcon-TherearemanyimplementationsofMapReducesuchcurrentdatamodifications.Tothisend,webuildMR-asMap-Reduce-Merge[6],MapReduceforMulti-coreBenchtodealwith

6、largevolumesofrelationaldataandandMultiprocessorsystems[11],GoogleMapReduce,executehighlycomplexqueries.ByMRBench,usersandApacheHadoop[1].canevaluatetheperformanceofMapReducesystemsEvenifMapReduceimplementationsarealreadybe-whilevaryingenvironmentalparameterssuchasdataingusedwidely,thereareon

7、lyafewwaystomeasuresizeandthenumberof(Map/Reduce)tasks.Ourex-performanceofMapReduceimplementations.TheretensiveexperimentalresultsshowthatMRBenchisaaresomeresearches[7,11,1]whichtriedtobench-usefultooltobenchmarkthecapabilityofansweringmarkMa

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。