欢迎来到天天文库
浏览记录
ID:40236510
大小:670.77 KB
页数:38页
时间:2019-07-27
《hadoop-数据挖掘研究组》由会员上传分享,免费在线阅读,更多相关内容在行业资料-天天文库。
1、HadoopIntroducingInstallationandConfiguration数据挖掘研究组DataMiningGroup@XiamenUniversityADistributeddata-intensiveProgrammingFrameworkHDFSMapReduceHadoopDistributedstorageParallelcomputing数据挖掘研究组DataMiningGroup@XiamenUniversityIntroducingtoHDFSHadoopDistributedFileSystem(HDFS)Anopen-sourceimp
2、lementationofGFShasmanysimilaritieswithdistributedfilesystems.However,comesdifferenceswithit.HDFSishighlyfault-tolerantandisdesignedtobedeployedonlow-costhardware.HDFSprovideshighthroughputaccesstoapplicationdataandissuitableforapplicationsthathavelargedatasets.数据挖掘研究组DataMiningGroup@Xiame
3、nUniversityHowitworks?FeaturesofitAnimportantfeatureofthedesign:dataisnevermovedthroughthenamenode.Instead,alldatatransferoccursdirectlybetweenclientsanddatanodes数据挖掘研究组DataMiningGroup@XiamenUniversityMapReduce?Let’stalkitnexttime………数据挖掘研究组DataMiningGroup@XiamenUniversity“RunningHadoop?”W
4、hatmeansforit?“RunningHadoop”meansrunningasetofdaemons.NameNodeDataNodeSecondaryNameNodeJobTrackerTaskTracker数据挖掘研究组DataMiningGroup@XiamenUniversityWhoWorksforwho?HDFSMapReduceHadoopNameNodeSecNDTaskTrackerJobTrackerDataNodeNameNodeHadoopemploysamaster/slavearchitectureforbothdistributeds
5、torageanddistributedcomputation.NameNodeisthemasterofHDFSthatdirectstheslaveDataNodedaemonstoperformthelow-levelI/OtasksNameNodeisthebookkeeperofHDFSkeepstrackofhowyourfilesarebrokendownintofileblockskeepstrackoftheoverallhealthofthedistributedfilesystemDataNodereadingandwritingHDFSblocksforc
6、lientscommunicatewithotherDataNodestoreplicateitsdatablocksforredundancy数据挖掘研究组DataMiningGroup@XiamenUniversityNameNodeandDataNodeSecondaryNameNodeSNNisanassistantdaemonformonitoringthestateoftheclusterHDFSdiffersfromtheNameNodeinthatthisprocessdoesn’treceiveorrecordanyreal-timechangestoHD
7、FScommunicateswiththeNameNodetotakesnapshotsoftheHDFSmetadataRecovery:NameNodefailure????WereconfiguretheclustertousetheSNNastheprimaryNameNodeJobTrackertheliaisonbetweenyourapplicationandHadoopsubmityourcodetoyourcluster,theJobTrackerdeterminestheexecutionplan
此文档下载收益归作者所有