Oracle iAS 10g 多层次高可用性体系构架

Oracle应用服务器10g的高可用性MarkMaHongWei内容Oracle应用服务器高可用性简述–HA的体系构架–新版本中的新功能–总体简述高可用性(包括中间层和基础结构层)基础结构的高可用性–冷故障切换和Active集群–用户身份管理数据复制–使用客户数据库的RAC功能中间层的高可用性–WebCache,OHS,OC4J–Portal,Forms,ReportsandotherMiddleTiers总结HA体系结构简述LoadBalancer/WebCacheWireless&MobileBrowserWebUsersRACMiddleTierClustersOHSwithLoadBalancingInfrastructureClusterClientOracleApplicationServer10g端到端的高可用性Oracle应用服务器高可用性系统故障数据故障和灾难人为错误Loadbalancers,Clusters,OPMNDisasterRecovery,Backup&RecoveryB&R,Configuration“flashback”系统管理数据管理RollingUpgradesShouldn’tbeany(dataindatabase)计划外宕机计划内宕机主要信息和定位网格计算中的主要功能无单点故障(End-to-EndHighAvailability):–CommonframeworkforDeathDetection,ProcessMonitoringandProcessRestart–InfrastructureClusters–MiddleTier:Loadbalancing,SmartRouting,–Data/Session-stateReplication(OC4J/WebCache/OID)–TransparentApplicationFailover(TAF)最小化计划内和计划外的宕机时间–RollingUpgrades–DisasterRecoveryforMTaswellasInfrastructure–BackupandRecovery–Configuration“undo”(Archival&RecoveryusingDCM)–Hotdeployment新版本中的新特性新的、智能的均衡负载/routingalgorithmsformod_oc4j基于文件系统的OC4Jclusters自主管理功能(OPMNdeathdetection&restart)支持OC4J应用的热部署提供快速的JDBC故障接管灾难恢复–forbothmiddletiersandinfrastructure备份和恢复–IncrementalBackupandRecoveryDCMArchival&Recovery–“Flashback”toapreviousconfigurationRollingUpgradesOracleAS中高可用的核心技术OPMN(Oracle进程管理和告知系统):–ProcessMonitoring(DeathdetectionandRe-start)–SelfHealing–AvailableforthefollowingOracleASprocesses:OHS,OC4J,WebCache,OID,Wireless,Reports,Discoverer,ProcessConnect,DCMDaemon,LogLoaderDCM(分布式配置管理):–File-basedrepositoryordatabase-basedrepository–Consistentconfigurationmanagement–helpspreventhumanerrorswhenconfiguringOC4Jclusters–AvailableforDCM-managedcomponents灾难恢复端到端的远程恢复方案Active/Standby配置Primary&standby节点必须使用同样的虚拟服务器名称Usesbackupandrecoverytosynchronizeprimary&standbysitesMT1MT2InfraClientsMT2*MT1*Infra*PrimarySiteStandbySiteDRSyncDRSyncDRSyncLoadBalancer灾难恢复MT1MT2InfraClientsMT2*MT1*Infra*PrimarySiteStandbySiteDRSyncDRSyncDRSyncLoadBalancer端到端的远程恢复方案Active/Standby配置Primary&standby节点必须使用同样的虚拟服务器名称Usesbackupandrecoverytosynchronizeprimary&standbysites基础结构的高可用性高可用性服务组件–IdentityManagementServices–forSSO/OID–MetadataRepositoryServices–forPortal,Wireless,Discoverer,etc.–ManagementServices–forEnterpriseManagerHA选项–ColdFailoverCluster–ActiveCluster–IdentityManagement(IM)Replication–InstallMetadataRepositoryinexistingRACDB冷故障接管的集群Active/Passivesetup:–Onenodeis“hot”andtheothernodeis“cold”SingleInstall–Onsharedstoragedevice–Usingvirtualhostname&IP(theclusterwareassociatesthemwiththeactivenode)Clustervendorssupported:–Sun,HP,IBM,RedHat,Veritas,Fujitsu-Siemens–FailSafeforWindowsMTMTActiveInfraClientsInactiveInfraSharedstorageHardwarecluster$ORACLE_HOME-OID-SSO-DAS-InfraDBinstanceV.Host冷故障接管的集群MTMTInactiveInfraClientsActiveInfraSharedstorageHardwarecluster$ORACLE_HOME-OID-SSO-DAS-InfraDBinstanceActive/Passivesetup:–Onenodeis“hot”andtheothernodeis“cold”SingleInstall–Onsharedstoragedevice–Usingvirtualhostname&IP(theclusterwareassociatesthemwiththeactivenode)Clustervendorssupported:–Sun,HP,IBM,RedHat,Veritas,Fujitsu-Siemens–FailSafeforWindowsV.HostActive集群MTMTClientsSharedstorageLoadBalancerActiveInfraNode1ActiveInfraNode2HardwareclusterInfraDBfilesLocalstorageLocalstorageConfigFiles($O_H2)ConfigFiles($O_H1)OID1SSO1DAS1DBinstance1OID2SSO2DAS2DBinstance2Active/Activesetup–AllnodesareactiveProvides“high9s”uptimeSingleInstall–DB(RAC)onshareddisk–Rest-of-infraonlocaldisks–RequiresloadbalancerLimitedReleasein9.0.4:–Sun,HP,Linux–Fullproductionversiontofollowinnextrelease(andonallplatforms)AlsoprovidesscalabilityActive集群MTMTClientsSharedstorageLoadBalancerActiveInfraNode1ActiveInfraNode2HardwareclusterInfraDBfilesLocalstorageLocalstorageConfigFiles($O_H2)ConfigFiles($O_H1)Active/Activesetup–AllnodesareactiveProvides“high9s”uptimeSingleInstall–DB(RAC)onshareddisk–Rest-of-infraonlocaldisks–RequiresloadbalancerLimitedReleasein9.0.4:–Sun,HP,Linux–Fullproductionversiontofollowinnextrelease(andonallplatforms)AlsoprovidesscalabilityOID1SSO1DAS1DBinstance1OID2SSO2DAS2DBinstance2用户身份管理的复制IdentityMgmtMTOID2DB1ConfigFiles($O_H1)OHS2OC4J2SSO2IdentityMgmtMTOHS1OC4J1SSO1LoadBalancer(HTTPtraffic)LoadBalancer(LDAPtraffic)OID1DB2ConfigFiles($O_H2)DataReplicationMTMTMTMTMultipleIMMiddleTiersMultipleOIDMultipleIMDBs(keptsynchronizedusingReplication)Idealforgeographicallydispersedsites中间层高可用性WebCache集群提供了冗余和容错OHS/mod_oc4jloadbalancingandsmartrouting–weighted,metrics-based,roundrobin,random–localaffinityFilebasedrepositoryforOC4JclustersEJBProxy:–Auto-reroutingofrequests交易应用的故障接管TAFofJDBCconnectionstoRACdatabase–SupportedwithJDBCThickdriver(OCIdriver)–Defineoracle.jdbc.OracleOCIFailoverInterfaceforstatefulapplications(tohandlestaterestore)Configurationmodes:–BasicMethod:reconnecttoavailableinstanceuponfailure–Pre-connectMethod:maintainbackupconnectiontoanotherRACinstanceforfasterTAF–FornonRACdatabaseauto-retrywithorwithouttimeoutOracle应用服务器的集群均衡负载–WebCache/LB–Mod_oc4j[Servlets]Scalemiddle-tier‘linearly’–LooselycoupledSession的复制克隆系统配置跨集群的应