Cassandra与HBase系统架构比对

Cassandra&HBaseComparison郭鹏盛大在线新浪微博：@逖靖寒SACC2011SACC2011FunctionImplementWhat’sNextAgendaSACC2011SACC2011FunctionImplementWhat’sNextAgendaSACC2011SACC2011DataModelKeyspaceRowColumnFamilyColumnNameColumnValueTimestampKS-1Row-1cfName-1colName-1colValue-112345678KS-1Row-1cfName-1colName-2colValue-222345638KS-1Row-1cfName-2colName-1colValue-112445678KS-1Row-2cfName-1colName-1colValue-112345678KS-2Row-1cfName-1colName-1colValue-112345678StandardColumnFamilySACC2011SACC2011DataModelKeyspaceRowColumnFamilySuperColumnNameColumnNameColumnValueTimestampKS-1Row-1cfName-1sColName-1colName-1colValue-112345678KS-1Row-1cfName-1sColName-1colName-2colValue-222345638KS-1Row-1cfName-2sColName-2colName-1colValue-112445678SuperColumnFamilySACC2011SACC2011DataModelTableRowColumnFamilyColumnNameColumnValueTimestampVersionTab-1Row-1cfName-1colName-1colValue-112345678V-1Tab-1Row-1cfName-1colName-2colValue-222345638V-2Tab-1Row-1cfName-1colName-2colValue-212445678V-1Tab-1Row-2cfName-1colName-1colValue-112345678V-1Tab-2Row-1cfName-1colName-1colValue-112345678V-1SACC2011SACC2011BasicAPICRUDConsistencyLevelONEQUORUMALLRangeQuery：RowTokenSACC2011SACC2011AdvancedAPIAuthorization&AuthenticationSecondaryIndexOnlineSchemaUpdateCounterMapReduceCQL(CassandraQueryLanguage)SACC2011SACC2011CRUDScan:RowBasicAPISACC2011SACC2011AdvancedAPILockPutcheckAndPutDeletecheckAndDeleteLimitedOnlineSchemaUpdateCounterMapReduceFilterSACC2011SACC2011FunctionImplementWhat’sNextAgendaSACC2011SACC2011SystemArchitectureCommitlogAPPCF-NRAMMemtableSSTableCommitlogAPPCF-1MemtableSSTableSystemCFLocalDiskCassandraServer1CassandraServerN…SACC2011SACC2011SystemArchitectureSACC2011SACC2011FileSystemSACC2011SACC2011ConsistencyServerNo.RowKeyListCassandra-01key1,key2,key3,key4Cassandra-02key2,key3,key4,key5Cassandra-03key3,key4,key5,key6ServerNo.RowKeyListRegionServer-01key1,key2,key3,key4RegionServer-02key5,key6,key7,key8RegionServer-03key9,key10,key11,key12SACC2011SACC2011LocationSACC2011SACC2011LocationSACC2011SACC2011StorageFileSACC2011SACC2011CacheCacheValueRowCacheKeyIndexKeyCacheKey/ValueCacheValueBlockCacheDataBlockSACC2011SACC2011BulkLoaderBinaryMemtableHRegionInterface.bulkLoadHFile•TableName•RegionName•CFNameLoadIncrementalHFiles•HFilePath•TableNamenodetoolcleanupsstableloaderSACC2011SACC2011LoadBalancenodetoolmovetokenbalancermoveregionNameserverNamesplitregionNameSACC2011SACC2011FunctionImplementWhat’sNextAgendaSACC2011SACC2011EnhancedHadoopandHivedistributionthatutilizesCassandraforallofitscoreservices.IntegratedwithHadoopMapReduceandHiveProvidinganHDFScompatiblestoragelayerpoweredbyCassandraBriskSACC2011SACC2011ApplicationStackSACC2011SACC2011DeploySACC2011SACC2011CoprocessorMasterObserverRegionObserverEndPointSACC2011SACC2011OpenFlushClose…PreOpenPostOpen……HMasterMasterObserver……SACC2011SACC2011RegionObserverGetPutScan…PreGetPostGet…………RegionServerSACC2011SACC2011EndPointDefineyourownextensionstoHBaseRPCtransactionsexchangedbetweenclientsandtheregionservers.SACC2011SACC2011HBaseaccesscontrol:providesbasicaccesscontrolforHBaseColumnaggregate:SQL-likesum(),avg(),max(),min(),etc.RegionlevelindexingHBasetable,regionaccessstatisticApplicationsSACC2011SACC2011KeyValueKeyScanColumn动态添加排序二级索引使用场景SACC2011SACC2011从零开始快速开发快速部署简单的安全认证Cassandra适用场景SACC2011SACC2011已有Hadoop集群要求数据一致性高效的数据平衡服务端代码执行HBase适用场景SACC2011SACC2011Thanks!SACC2011SACC2011