VMware容灾解决方案

©2009VMwareInc.AllrightsreservedVMware容灾解决方案曹连雨[e-mail]传统灾难恢复面临的挑战昂贵复杂的恢复计划????????不可靠的故障切换AppsHostsStorageNetworkSoftwareHostsStorageFacilities$10Kperapp难以满足业务需求过长RTOs–几天到几周消耗过多时间和资源vSphere为灾难恢复提供了坚实基础FlexibleInfrastructureEliminateneedforidenticalhardwareacrosssitesEnablewaterfallingofequipmenttorecoverysiteSimpleApplicationProtectionEntiresystem–includingapplication,OS,anddata–isstoredasvirtualmachinefilesEntiresystemcanbeprotectedwithdataprotectiontoolsCost-EfficientInfrastructureReducedhardwarerequirementsatrecoverysiteUserecoveryhardwaretorunlow-priorityapps封装整合硬件无关vSpherevSpherevSpherePlannedDowntimeUnplannedDowntimeVMware在各个层面提升业务连续性•vMotion•StoragevMotion•FaultTolerance•HighAvailability•AppHA•SiteRecoveryManager•DRtotheCloudwithSRM•vSphereReplication•vSphereDataProtectionAdvanced•vSphereAPIsforDataProtection(VADP)SiteApplicationAvailabilityLocalApplicationAvailabilityDataProtectionVMwareSRM实现简单、可靠的灾难恢复SiteRecoveryManager作为vSphere的补充，可为所有应用程序提供最简单可靠的灾难防护并实现站点迁移简化恢复和迁移计划的管理工作使用集中式灾难恢复计划取代人工操作手册将设置新计划所需的时间从数周缩短至数分钟自动执行故障切换和迁移流程，以确保可靠的恢复支持频繁的无中断测试确保快速、自动化的故障切换支持多种数据复制技术内置vSphereReplication为基于存储的复制提供广泛支持VMwarevSphereVMwarevCenterServerSiteRecoveryManagerVMwarevCenterServerSiteRecoveryManagerVMwarevSphereSiteA(Primary)SiteB(Recovery)ServersServersSRM简化恢复计划的创建和管理WeeksormonthstosetupError-proneQuicklyfallsoutofsyncwithappsandinfrastructurechangesSimplerecoveryplansetupinminutesFewerstepsmeansfarlessroomforerrorsSimpletokeepinsyncwithchanges…toSimpleRecoveryPlansFromComplexRunbooks…SRM容灾计划测试降低故障恢复风险SRMprovidesassurancethatDRobjectiveswillbemetTimeDRTestDRTestTESTINGGAPRecoveryRiskTraditionalDisasterRecoveryRecoveryRiskDRTestDRTestTimeSiteRecoveryManagerFrequentDRTestingSRM提供多种复制技术选项vSphereReplicationSimple,cost-efficientreplicationforTier2applicationsandsmallersites（RTO15m-24h；异步复制；Hostoverhead；Windows应用一致性）Storage-basedReplicationHigh-performancereplicationforbusiness-criticalapplicationsinlargersites（EMCRecoverPoint、SRDF、MirrorView、CelerraReplicator）vCenterServerSiteRecoveryManagervSpherevSphereReplicationStorage-basedreplicationvCenterServerSiteRecoveryManagervSphereSiteA(Primary)SiteB(Recovery)站点自动灾难切换OverviewBenefitsAutomaticallydetectsitefailuresRequireusertomanuallyinitiatefailoverAutomaterecoveryprocessStopreplicationandpresentreplicatedLUNstovSphereExecuteuser-definedrecoveryplanEnsurefastandpredictablefailoversandmigrationsConsistentlymeetbusinessrequirementsMinimizeriskofusererrorsSiteBSiteAReplication1Raisealertwhenheartbeatlost2Userinitiatesfailover3StopreplicationandpresentLUNstovSphere4RecoverVMsDRFailovervSpherevSphereSimplifyfailbackprocessAutomatereplicationmanagementEliminateneedtosetupnewrecoveryplanStreamlinefrequentbi-directionalmigarations自动灾备计划回滚实现双向容灾Re-protectVMsfromSiteBtoSiteAReversereplicationApplyreverseresourcemappingAutomatefailoverfromSiteBtoSiteAReverseoriginalrecoveryplanRestrictionsDoesnotapplyifSiteAhasundergonemajorchanges/beenrebuiltNotavailablewithvSphereReplicationOverviewBenefitsAutomatedFailbackSiteBSiteAReverseReplicationReverseoriginalrecoveryplanvSpherevSphere计划内站点迁移确保应用一致性和零数据丢失OverviewBenefitsTwoworkflowscanbeappliedtorecoveryplans:DRfailoverPlannedmigrationPlannedmigrationensuresapplicationconsistencyandnodata-lossduringmigrationGracefulshutdownofproductionVMsinapplicationconsistentstateDatasynctocompletereplicationofVMsRecoverfullyreplicatedVMsBettersupportforplannedmigrationsNolossofdataduringmigrationprocessRecover‘application-consistent’VMsatrecoverysitePlannedMigrationSiteBSiteAReplication1ShutdownproductionVMs2Syncdata,stopreplicationandpresentLUNstovSphere3Recoverapp-consistentVMsvSpherevSphere参考架构一：OracleDataGuard与SRM集成方案vCenterServerSiteRecoveryManagervSpherevCenterServerSiteRecoveryManagervSpherevSphereReplicationSiteA(Primary)SiteB(Recovery)PrimarySAPDBStandbySAPDBOracleDataGuardLogShippingSAPCSSAPPASSAPCSSAPPAS优点：应用级数据一致性。参考架构二：EMCRecoverPoint与SRM集成方案VMwareSRM普通接口各种存储设备广域网SAN远程副本SANSANRecoverPoint生产LUNS无干扰的DR测试虚拟机自动接管RecoverPointEMCLUNLUNLUNSRM服务被保护的虚拟机（SQLServer、Exchange、Oracle）生产端容灾端本地副本优点：应用级数据一致性；高效复制技术；同步复制。参考架构三：Rackspace高可用云基础架构IP或FC故障域B故障域A多达4个引擎引擎1引擎2引擎3引擎4多达4个引擎引擎1引擎2引擎3引擎4扩展的主机群集VPLEX构建双活数据中心实现业务连续可用性全局缓存吻合性虚拟化活动I/O到和从任何通道分布式写（通过WAN）/本地化读即使一个阵列出现完全失败也不中断异构异构通过VPLEX镜像完全容忍一到多个控制器故障分布式虚拟卷–全局可访问主机主机用于站点故障恢复HA示例扩展的VSPHERE群集(DRS+HA)IPIP站点A站点BAAAAAAAAAAVMs自动重启在这个站点的所有VMs都将失败AccessAnywhere™由VPLEXWitness指导，站点B保持在线VPLEXWITNESSVMWAN100%分布式DRS示例（负载均衡）扩展的VSPHERE群集(DRS+HA)站点A站点BAAAAAAAAAA平衡池中的VMsAccessAnywhere™WAN0%50%50%参考架构四：VPLEX+RecoverPoint:两地三中心架构采用VPLEXMetro同步技术，扩展vSphere集群，两个数据中心实现active-active的高可用通过RecoverPoint的CRR数据复制技术实现第三点的数据容灾可以从第三点对双活数据中心进行逻辑数据恢复FederatedAcessAnywhereVPLEXWITNESSSyncDistanceAsyncDistanceDRCopyVPLEXRecoverPointSiteB扩展的VSPHERE群集(DRS+HA)SiteAvSphereDRSiteVMwareSiteRecoveryManagerVPLEXRecoverPointUsingVPLEXMetrowithVMwareHA-vMotionRequirements1.最大的FC网络的roundtrip的latency不可以超过10ms2.最小的IP网络带宽需求是622Mbps,建议带宽至少1Gbps，若vMotion频繁建议升级为10Gbps,vSphere5.0支持多网卡vMotion以提升效率。3.兩端的数据中心的ES