Oracle数据库集群和分区技术.pptx

讲座议题讲座议题为什么要使用为什么要使用 Oracle RAC Oracle RAC 数据库数据库 RAC 的硬、软件体系架构的硬、软件体系架构 RAC 的工作原理和的工作原理和Cache Fusion 均衡负载和高可用性均衡负载和高可用性 Transparent Application Failover第1页/共68页为什么要使用为什么要使用 Oracle RAC 数据库数据库?稳定的、稳定的、24*7的高可靠性系统的高可靠性系统均衡负载和强大的处理能力均衡负载和强大的处理能力灵活开放的扩展能力灵活开放的扩展能力随用随付、降低采购和维护成本随用随付、降低采购和维护成本第2页/共68页采用集群数据库技术采用集群数据库技术Trans-Pacific FlightRock ConcertThis:Or this:This:Or this:第3页/共68页Computer BComputer BComputer BComputer AComputer AComputer A操作系统级的操作系统级的HA主备机模式主备机模式高可用性：主机运行，备机等待主机故障、备机接管接管磁盘阵列控制IP地址自动漂移重新在备机启动数据库重新连接应用程序实施简单缺点：切换间存在停顿备机空闲第4页/共68页提供提供24*7 的安全可靠性的安全可靠性透明的应用程序故障切换、消除单点故障透明的应用程序故障切换、消除单点故障 应用程序和用户连接在故障发生时，可以自动透明地应用程序和用户连接在故障发生时，可以自动透明地连接到其它的服务器上连接到其它的服务器上 应用程序和查询不会中断，用户感觉不到故障的发生应用程序和查询不会中断，用户感觉不到故障的发生 安全认证信息可持续保留安全认证信息可持续保留ComputerComputerA AComputerComputerB BComputerComputerA AComputerComputerB BNode A in acluster fails,users are migrated第5页/共68页“冷冷”故障接管的限制故障接管的限制冷接管速度慢移动硬盘卷、mount 上逻辑卷、启动 Oracle 实例，打开数据文件不能够预先连接到备份 Server 上出现故障后，所有的会话都必须重新连接到数据库实例上冷的数据库缓存接管后数据库的缓存是“冷”的，减低数据库的性能不能够很容易的扩展第6页/共68页Oracle10g RAC 128节点的网格数据库第7页/共68页自动动态负载均衡实现可伸缩性自动动态负载均衡实现可伸缩性l数据库启动时，向监听进程注册l节点向监听进程报告 CPU 的使用情况l当建立连接时，监听进程选择最空闲的节点处理请求NodeDispatchersNodeDispatchersListenerListenerCPU LoadCPU Load第8页/共68页灵活的可伸缩性和强大的处理能力灵活的可伸缩性和强大的处理能力第9页/共68页讲座议题讲座议题为什么要使用为什么要使用 Oracle RAC Oracle RAC 数据库数据库 RAC 的硬、软件体系架构的硬、软件体系架构 RAC 的工作原理和的工作原理和Cache Fusion 均衡负载和高可用性均衡负载和高可用性 Transparent Application Failover第10页/共68页Oracle RAC集群数据库技术硬件结构集群数据库技术硬件结构ClusteredDatabase InstancesMirrored Disk SubsystemHigh Speed Switch or InterconnectHub or Switch FabricApplication Servers/NetworkCentralized Management ConsoleStorage Area NetworkLow Latency InterconnectDrive and Exploit Industry Advances in ClusteringUsersNo SinglePoint Of FailureShared CacheScalable&throughput第11页/共68页Cluster 硬件结构硬件结构节点(Node)内部连接(Interconnect)共享磁盘(Shared disk)第12页/共68页节点节点(Nodes)自包容的服务器，单 CPU 或者多 CPU 系统(SMP/NUMA)一个集群需要两个或者更多的节点每个节点支持一个 RAC Instance每个节点可以支持不同类型的应用每个节点不要求配置完全相同，但建议同配置节点可以动态的添加和删除Node第13页/共68页内部连接内部连接(Interconnect)连接节点之间，节点间通讯的高速通道可以使用标准的网络协议，如：TCP/IP使用高速的连接协议(high speed interconnects)可以获得更好的性能UDPVIA厂商专有的，经过验证的协议第14页/共68页网络配置Cluster InterconnectPublic LANOptionalCluster InterconnectPublic LANWorkstationNODE2NODE3NODE1Optional 每个节点需三个IP地址一个公共服务地址(提供客户访问)一个私有的内部地址(仅内部使用)一个虚拟IP地址(与公共服务地址同一网段)第15页/共68页共享磁盘系统共享磁盘系统(Shared Disk)共享磁盘上的共享的设备 传统上的裸设备卷(UNIX)逻辑盘(Windows)Oracle OCFS，ASM集群文件系统厂商专有的，经过验证的系统，如Veritas，GPFS共享磁盘系统第16页/共68页Oracle9i RAC Architecturepublic networkoperating SystemclusterwareOracle 9i RACinstance 1server 1shared storageredo log instance 1 redo log instance 3 control filesdatabase filesclusterinterconnectcache to cacheDB cacheserver 2operating systemclusterwareOracle 9i RACinstance 2DB cachesrvcfg/var/opt/oracle/srvConfig.locquorumServer 3,4,.第17页/共68页Oracle 10g RAC Architecturepublic networkoperating SystemCRSOracle 10g RACinstance 1server 1shared storageredo log instance 1 redo log instance 3 control filesdatabase filesserver 2operating systemCRSOracle 10g RACInstance 2clusterinterconnectcache to cacheDB cacheDB cacheocr/var/opt/oracle/ocr.locvoting/var/opt/oracle/ocr.locServer 3,4,.第18页/共68页Oracle RAC 系统管理工具系统管理工具The Server Control Utility(SRVCTL)管理整个 Cluster 环境 为其他管理工具提供对 RAC 的管理基于 GSD 基础之上Node 2GSDSRVCTLAgentNode 1GSDSRVCTLAgent第19页/共68页讲座议题讲座议题为什么要使用为什么要使用 Oracle RAC Oracle RAC 数据库数据库 RAC 的硬、软件体系架构的硬、软件体系架构 RAC 的工作原理和的工作原理和Cache Fusion 均衡负载和高可用性均衡负载和高可用性 Transparent Application Failover第20页/共68页Scaleup and SpeedupOriginal systemHard-ware100%of taskTimeCluster system scaleupup to 200%oftaskup to 300%oftaskHard-wareTimeHard-wareTimeHard-wareTime50%of taskCluster system speedupHard-wareTimeHard-wareTime50%of task第21页/共68页Speedup and Scaleup for Different Types of WorkloadsWorkloadOLTP and InternetDSS with parallel queryBatch(mixed)SpeedupNoYesPossibleScaleupYesYesYes第22页/共68页Cache FusionCache Fusion 提供提供 Oracle 集群数据库透明的可扩展性集群数据库透明的可扩展性Cash Fusion 机制使机制使 Oracle Instance 间可以直接传递数据块间可以直接传递数据块Cache Fusion 能够跟踪资源的位置和状态能够跟踪资源的位置和状态Global Cache Service 管理整个集群数据库的缓存资源管理整个集群数据库的缓存资源Resource Master(Global Resource Directory)分布在活动的分布在活动的 Instance 中，并可以动态的中，并可以动态的 remaster第23页/共68页Cache FusionTM 体系结构在节点A对数据进行了修改并提交，在节点B对同一数据块进行读操作，可通过高速链路访问多版本数据缓存支持节点A继续对数据块进行修改DatabaseNode ADatabase buffersNode BDatabase buffers两节点间的数据传输直接通过内部高速通道实现，大大减少了硬盘I/O数据传输Select第24页/共68页Cache FusionTM 提高了效率20 msec.01 msec100 msec本地内存的数据远程节点的数据 硬盘上的数据1 msec100100 x xOracle8Oracle8Oracle9Oracle9i i第25页/共68页Global Resource DirectoryExampleInstance ANode 1Global ResourcesID20232629Grantedto InstanceA,B,CA,BCB,CInstance BNode 2Global ResourcesID21242730Grantedto InstanceBA,CA,B,CB,CInstance CNode 3Global ResourcesID22252831Grantedto InstanceA,CCA,B,CB,C第26页/共68页全局缓存服务资源模式全局缓存服务资源模式缓存资源采用以下三种模式之一缓存资源采用以下三种模式之一:NULL Shared(S)Exclusive(X)第27页/共68页缓存融合举例缓存融合举例n第28页/共68页Cache Fusion Block Transfers:Example OverviewInstance A1008Instance DResourcemasterInstance CInstance B第29页/共68页Example 1:Read with No TransferInstance A1008Instance DResourcemasterInstance CInstance BRequest toobtain a shared resource on CN1第30页/共68页Example 1:Read with No TransferInstance A1008Instance DResourcemasterInstance CInstance BThe request is granted and the requesting instance is informedNS12第31页/共68页Example 1:Read with No TransferInstance A1008Instance DResourcemasterInstance CInstance BSRead request132第32页/共68页Example 1:Read with No TransferInstance A1008Instance DResourcemasterInstance CInstance BSBlock imagedelivered10081234第33页/共68页Example 2:Read to Write TransferInstance A1008Instance DResourcemasterInstance CInstance BRequest to obtain an exclusive resource on BS1008N1第34页/共68页Example 2:Read to Write TransferInstance A1008Instance DResourcemasterInstance CInstance BInstruction to transferthe block to B for exclusive accessS1008N21第35页/共68页Example 2:Read to Write TransferInstance A1008Instance DResourcemasterInstance CInstance BBlock and resource status(including Cs plan to close its resource)SN1008N1008213第36页/共68页Example 2:Read to Write TransferInstance A1008Instance DResourcemasterInstance CInstance BResource assumptionand status messageN1008NX10092134第37页/共68页Example 3:Write to Write TransferInstance A1008Instance DResourcemasterInstance CInstance BNX1009Request to obtain resource inexclusive mode1第38页/共68页Example 3:Write to Write TransferInstance A1008Instance CInstance BNX1009Instruction to transferexclusive resource to B12Instance DResourcemaster第39页/共68页Example 3:Write to Write TransferInstance A1008Instance DResourcemasterInstance CInstance BNXN10091009Exclusive-keep copy of buffer123第40页/共68页Example 3:Write to Write TransferInstance A1008Instance DResourcemasterInstance CInstance BNX10091013N1234第41页/共68页Example 4:Write to Read TransferInstance A1008Instance DResourcemasterInstance CInstance BX10091013NNRequest to obtainresource in shared mode1第42页/共68页Example 4:Write to Read TransferInstance A1008Instance DResourcemasterInstance CInstance BX10091013NNInstruction to transfer shared resource to C12第43页/共68页Example 4:Write to Read TransferInstance A1008Instance DResourcemasterInstance CInstance BXS10091013NN1013Shared-keep copy of buffer321第44页/共68页Example 4:Write to Read TransferInstance A1008Instance DResourcemasterInstance CInstance BS10091013NNSResource assumptioninformation10132341第45页/共68页讲座议题讲座议题为什么要使用为什么要使用 Oracle RAC Oracle RAC 数据库数据库 RAC 的硬、软件体系架构的硬、软件体系架构 RAC 的工作原理和的工作原理和Cache Fusion 均衡负载和高可用性均衡负载和高可用性 Transparent Application Failover第46页/共68页Load Balancing with Oracle Net ServicesInstancesDatabaseListenersClients第47页/共68页Client Load BalancingClientsListeners(DESCRIPTION=(ADDRESS_LIST=(LOAD_BALANCE=on)(ADDRESS=(PROTOCOL=tcp)(HOST=sales1)(PORT=1521)(ADDRESS=(PROTOCOL=tcp)(HOST=sales2)(PORT=1521)(CONNECT_DATA=(SERVICE_NAME=第48页/共68页Connection Load Balancingnode1_lsnr=(DESCRIPTION=(ADDRESS_LIST=(ADDRESS=(PROTOCOL=tcp)(HOST=node1)(PORT=1580)node2_lsnr=(DESCRIPTION=(ADDRESS_LIST=(ADDRESS=(PROTOCOL=tcp)(HOST=node2)(PORT=1580)Parameter file on node1local_listener=node1_lsnrremote_listener=node2_lsnr第49页/共68页Connection Load BalancingNodes(DESCRIPTION=(ADDRESS_LIST=(LOAD_BALANCE=on)(ADDRESS=.)(ADDRESS=.)(ADDRESS=.)(CONNECT_DATA=(SERVICE_NAME=(SERVER=shared)Dispatchers第50页/共68页Service and Instance Names(DESCRIPTION=(LOAD_BALANCE=ON)(ADDRESS=(PROTOCOL=tcp)(HOST=host1)(PORT=1521)(ADDRESS=(PROTOCOL=tcp)(HOST=host2)(PORT=1521)(ADDRESS=(PROTOCOL=tcp)(HOST=host3)(PORT=1521)(DESCRIPTION=(ADDRESS=(PROTOCOL=tcp)(HOST=host1)(PORT=1521)(CONNECT_DATA=(INSTANCE_NAME=S1)第51页/共68页Adaptive Parallel QueryQuerycoordinatorParallel queryexecutionQuery processeshave node affinity for query coordinatorbut will use other nodes if neededNode 2Node 3Node 1第52页/共68页ListenersMultiple listeners enableClient load balancingConnect-time failoverTransparent Application Failover(TAF)ClientsCluster nodesListeners第53页/共68页Virtual IP Addresses and RAC第54页/共68页Primary and Secondary InstancesClientsNodesListeners.*.active_instance_count=1*.cluster_database_instances=2sales1.instance_name=sales1sales2.instance_name=sales2.Primary instance:first one tostartSecondary instance:second one to start;can become primary if other instance fails 第55页/共68页ExampleRAC1=(DESCRIPTION=(LOAD_BALANCE=OFF)(FAILOVER=ON)(ADDRESS=(PROTOCOL=TCP)(HOST=aaacme1)(PORT=1521)(ADDRESS=(PROTOCOL=TCP)(HOST=aaacme2)(PORT=1521)(INSTANCE_ROLE=PRIMARY)(SERVER=DEDICATED)(FAILOVER_MODE=(BACKUP=RAC2)(TYPE=SESSION)(METHOD=PRECONNECT)(RETRIES=180)(DELAY=5)第56页/共68页ExampleRAC2=(DESCRIPTION=(LOAD_BALANCE=OFF)(FAILOVER=ON)(ADDRESS=(PROTOCOL=TCP)(HOST=aaacme2)(PORT=1521)(ADDRESS=(PROTOCOL=TCP)(HOST=aaacme1)(PORT=1521)(INSTANCE_ROLE=SECONDARY)(SERVER=DEDICATED)(FAILOVER_MODE=(BACKUP=RAC1)(TYPE=SESSION)(METHOD=PRECONNECT)(RETRIES=180)(DELAY=5)第57页/共68页TAF in Real Application ClustersCan be used with spare nodes or with primary/secondary instance configurationsDesigned for this environment,but can be used for Real Application Clusters GuardReplicated systemsData Guard第58页/共68页TAF Failover Mode OptionsMust add failover options manually to TNS configuration filesThey are part of the CONNECT_DATA section of a connect descriptorFailover options includeTYPE:Identify the nature of TAF,if anyMETHOD:Configure how quickly failover can occurBACKUP:Identify an alternate net service nameRETRIES:Limit the number of times a reconnection will be attemptedDELAY:Specify how long to wait between reconnection attempts 第59页/共68页Failover types identify the nature of TAF,if anyThe options are:SESSION:Failover to an alternate session onlySELECT:Failover and continue with any ongoing queryNONE:Prevent failoverFailover Types.(CONNECT_DATA=(FAILOVER_MODE=(TYPE=SELECT).第60页/共68页Failover MethodsDetermine how quickly connections become available following a failoverBASIC:Establishes no contact with the failover instance prior to failurePRECONNECT:Creates mirror connections on the standby instance for the connections on the primary instance.(CONNECT_DATA=(FAILOVER_MODE=(METHOD=PRECONNECT).第61页/共68页Failover Backup ServiceSpecifies the net service name where failed connections will be directedAllows you to identify a specific backup instance connect descriptor for each primary instanceUse with the PRECONNECT method to identify the service to initiate your failover preconnections.(CONNECT_DATA=.第62页/共68页Failover Connection Reattempt OptionsRETRIES:The number of times to attempt to connect after a failoverDELAY:The amount of time in seconds to wait between connection attemptsSet the values based on your failover strategy and test results.(CONNECT_DATA=(RETRIES=20)(DELAY=15).第63页/共68页TAF Configuration ExampleHH=(DESCRIPTION=(ADDRESS_LIST=(LOAD_BALANCE=yes)(FAILOVER=true)(CONNECT_DATA=(SERVICE_NAME=hh)(failover_mode=(type=select)(method=basic)(retries=20)(delay=15)第64页/共68页Oracle Call InterfaceTAF 支持的程序接口：支持的程序接口：OCI programsJava JDBC thick drivers(OCI drivers)ODBC connectionsSQL*PlusSelect statements第65页/共68页AQ&问答问答 第66页/共68页第67页/共68页感谢您的观看。第68页/共68页