加入VIP
  • 专属下载特权
  • 现金文档折扣购买
  • VIP免费专区
  • 千万文档免费下载

上传资料

关闭

关闭

关闭

封号提示

内容

首页 IBM_PowerVM虚拟化介绍及案例经验

IBM_PowerVM虚拟化介绍及案例经验.pdf

IBM_PowerVM虚拟化介绍及案例经验

mountain100
2011-10-24 0人阅读 举报 0 0 暂无简介

简介:本文档为《IBM_PowerVM虚拟化介绍及案例经验pdf》,可适用于IT/计算机领域

IBMPowerVM虚拟化介绍及案例经验分享王泷高级系统工程师IBM系统科技事业部IBMPowerSystems演讲内容¾PowerVM虚拟化解决方案简介¾业界虚拟化技术比较¾PowerVM虚拟化技术–微分区和虚拟IO¾Power系统虚拟化新功能¾PowerVM在国内企业的成功案例IBMPowerSystemsVirtualizationwithoutLimitsDriveoverutilizationDynamicallyscaleperdemandDynamicEnergyOptimizationenergycostreductionEnergyScale™technologiesResiliencywithoutDowntimeRoadmaptocontinuousavailabilityHighavailabilitysystemsandscalingManagementwithAutomationVMControltomanagevirtualizationAutomationtoreducetasktimeWorkloadOptimizingSystemsAIXTheFutureofUNIXTotalIntegrationwithiScalableLinuxreadyforxConsolidationSmarterSystemsforaSmarterPlanetPowerVM:FoundationofthePowerSystemsSoftwareStackIBMPowerSystemsIBMdevelopshypervisorthatwouldbecomeVMonthemainframeIBMannouncesfirstmachinestodophysicalpartitioningIBMannouncesLPARonthemainframeIBMannouncesLPARonPOWERIBMintro’sPOWERHypervisor™forSystempandSystemi™IBMannouncesPowerVMIBMannouncesPOWER™,thefirstUNIXserverswithLivePartitionMobilityIBM虚拟化技术的领导者AyeartrackrecordinvirtualizationinnovationcontinueswithPowerVM™采用IBM虚拟化技术实现服务器集中的TCO分析节省机房空间、空调、用电等。运营成本更少的服务器数量和简化的IT架构自动化的管理工具人工更少的服务器数量以及更少的CPU数量显著降低软件的许可费用、升级和服务费用软件服务器数量的减少降低了维护所需费用维护成本分析预计节省费用IBMPowerSystemsy降低IT构建成本y计划内停机不再有应用中断y减少宕机机制y多服务器资源管理模式y更佳的能源利用y减少管理服务器数量y资源的动态调配yIO资源共享y更好的硬件资源利用率y减少管理服务器数量y资源灵活变更y更好的硬件资源利用率客户的获益y分区在线迁移yWPAR移动y动态内存共享y多CPU享池yVIOS:NPIV,IVEy虚拟配置管理y电源管理y安全管理y微分区y最小分区:CPUy虚拟IO服务器(VIOS)yVIOS映射y最高支持个分区y工作负载分区管理(PLM)y虚拟服务器(LPARs)y专属资源y最小分区:CPUy最高支持个分区y分区动态调整yIO划分yCUod功能新技术POWERPOWERPOWERƒ自从年起Systemp系列服务器在虚拟化方面一直保持领先地位ƒ客户已经从虚拟化方面获得了巨大的收益y繁多的物理服务器y服务器容量固定y大量的维护成本年以前领先的Power虚拟化技术IBMPowerSystemsIBMPowerVMVirtualization功能ƒProcessor–SharedordedicatedLPARs–CappedoruncappedLPARs–Multiplesharedprocessorpools–DynamicLPARoperations(addremove)–ShareddedicatedLPARsƒIO–SharedandordedicatedIO–VirtualEthernet,virtualSCSI–DynamicLPARoperations(addremove)–IntegratedVirtualEthernet–VirtualFC(NPortIDVirtualization)–VirtualTapeSupportƒMemory–Dedicatedmemory–ActiveMemorySharing–DynamicLPARoperations(addremove)–ActiveMemoryExpansion(POWERAIX)ƒOther–IntegratedVirtualizationManager–LiveLPARmobility–Workloadpartitions(AIX)–Workloadpartitionmobility(AIX)–LxforLinuxapplications(Linux)–IBMiVirtualization(IBMi)SharedProcessorPoolOSPowerVMHypervisorOSDedicatedProcessorLPARsLPARLPARWPARLPARSubPoolAOSOSLPARLPARSubPoolBOSOSLPARLPARLPARVirtualIOServerLPARIBMPowerSystems共享专有分区AIXLinuxiAIXLinuxiAIXLinuxi客户分区支持动态内存共享Lx分区在线迁移VMControlIVMDLPARSVIOSperServerPowerBlade,,Express多CPU共享池虚拟IO服务器VMControlIVM,HMCVMControlIVM,HMC管理平台Core(upto)Core(upto)最大分区数PowerBladePowerSystem*PowerBlade,PowerSystem*服务器EnterpriseStandardPowerVMEditionPowerVM的三种版本ƒPowerVMExpressEdition–Evaluations,pilots,PoCs–SingleserverprojectsƒPowerVMStandardEdition–Productiondeployments–ServerconsolidationƒPowerVMEnterpriseEdition–Multiserverdeployments–CloudinfrastructurePowerVMEditionsofferaunifiedvirtualizationsolutionforanyPowerworkloadsIBMPowerSystemsWebBrowserIVMLinuxCoresAIXVCoresPowerSystems应用PowerVM虚拟化技术VirtualIOserverƒSharedEthernetƒSharedSCSIFibreChannelattacheddisksubsystemsAIXVDynamicallyResizableCoresAIXVCoresCoresCoresLinuxSharedPoolMicroPartitioningFeatureƒShareprocessorsacrossmultiplepartitionsƒMinimumpartitionthcoreƒpartitionmaximumƒAIXV,Linux,IBMiManagedviaHMCorIVMVirtualIOpathsVirtualLANLinuxAIXVAIXVIBMiAIXVAIXVLinuxMicropartitioningNetworkNetworkLinuxEthernetSharingStorageSharingIntVirtManagerVirtualIOServerPartitionPOWERHypervisorLinuxCoresIBMiIBMPowerSystemsSoftwarerunsonahostoperatingsystemVMwareGSXMicrosoftVirtualServerHPIntegrityVMUserModeLinuxLinuxKVMSSI>PPPP>SI,SunDomains,HPnPartitionsBoardlevelpartitioningOriginalPOWERLPARHPvPartitionsSunLogicalDomainsAdjustablepartitionsPartitionControllerSMPServerOSAppsOSAppsHypervisorSMPServerOSAppsOSAppsHostOSSMPServerHypervisorOSAppsOSAppsHardwarePartitioningBareMetalHypervisorHostedHypervisor服务器虚拟化趋势ServerissubdividedintofractionseachofwhichcanrunanOSHypervisorprovidesfinegrainedtimesharingofallresourcesUsesOSservicestodotimesharingofallresourcesHypervisorfirmwarerunsdirectlyonserverSystemzPRSMandzVMPowerVMhypervisorVMwareESXServerXenHypervisorMicrosoftHyperVCorethreadlevelpartitioningIBMPowerSystemsNetworksandnetworkstorageUnassignedondemandresourcesHypervisorServiceProcessorProcessorsMemoryLinuxpartitionsHardwareManagementConsole*(HMC)AIXVExpansionslotsAIXVpartitionsVirtualStorageVirtualIOServerLocaldevicesandstorageWorkloadmanagementandprovisioningVirtualNetworksVirtualProcessors*IntegratedVirtualizationManager(IVM)isdisabledifHMCattached**AvailableonSystemp™QandbelowaswellastheBladeCenterJS,JSPower系统虚拟化架构AIXVpartitionsVEnetVSCSIIVM**LinuxKernelsAIXKernelsVirtualNetworkStorageIBMPowerSystemsIBMPowerVMEditionsvsSunLogicalDomainsNoNo‘Warm’migrationwithmanyconstraintsAddremoveCPUAddonlyvirtualIO(T)(T)SunUltraSPARCTTserversonlySunLogicalDomainsConsolidatemoreserverworkloadsMaximumPartitionsPowerVMcanconsolidateworkloadsfromfourdifferentplatformsYes(Lx)CrossplatformvirtualizationSuperiorperformanceforIOintensiveworkloadsYesSupportfordedicatedIODecreasedowntimeinyourenterpriseYesSupportforlivepartitionmobilityYesAllPowerSystemsserversPowerVMEditionsAdapttochangeswithoutdowntimeDynamiclogicalpartitioningStandardizeononevirtualizationsolutionacrossanentirelineofserversSupportacrossentireproductlinePowerVMBusinessBenefitVirtualizationcapabilityIBMPowerSystemsIBMPowerVMEditionsvsHPIntegrityVMYes(Lx)YesYesEALCertifiedYesCPUsTBRAMPowerVMEditionsHighlyflexibleconfiguration,withautomaticdeploymentofresourceswhereneededNoDynamicCPUsharingAdapttochangeswithoutdowntimeVMrebootDynamiclogicalpartitioningPowerVMiscapableofconsolidatingworkloadscompiledforfourdifferentplatformsNoCrossplatformvirtualizationSuperiorperformanceforIOintensiveworkloadsNoSupportfordedicatedIOSecuredenvironmentformissioncriticalapplicationsNoCertificationSecurityfaultisolationHelpsimproveTCOfromconsolidatingmoreworkloadsCPUs(max)GBRAMPartitionscalabilityPowerVMBusinessBenefitHPIntegrityVMVirtualizationcapabilityIBMPowerSystemsVMwareESXPowerVMBusinessValueScalabilityFactorsVirtualCPUsperVMMaximzeuseofyourlargest,mostexpensivesystemsbyvirtualizingevenlargeapplicationsincludingproductionMemoryperVMGBGBVirtualNICsperVMCPUsperphysicalserverLowercostbyincreasingutilizationbyconsolidatingtofewersystemsMemoryperphysicalserverGBGBVMsperphysicalserverFlexibilityFactorsDynamicvirtualCPUchangesinVMAdd(butnotRemove)YesMeetSLAswithoutdisruptionevenwhentheapplicationrequirementsarechangingDynamicmemorychangesinVMAdd(butnotRemove)YesDynamicIOdevicechangesinVMNoYesDirectaccesstoIOdevicesfromwithinVMSome(withNehalem)YesMaximumsimultaneouslivemigrationsMinimizeapplicationdowntimeManagementFactorsCapabilitytomanagePowerVM,VMWare,Xen,MicrosoftHyperVzVMNoYesusingIBMSystemsDirectorReducethecostofmanagingvirtulizationIBMPowerVMEditionsvsVMwareESXIBMPowerSystemsPowerVMVSVMwarePower的SAP虚拟化性能衰减对比IBMPowerSystemsPowerVMVSVMware双插槽至强服务器的SAP虚拟化性能衰减对比IBMPowerSystems用户购买了超大的服务器,并没有充分使用各种资源和功能CPUTotaljsptbank::::::::::::::::::::::::::::::::::::::::::::::::UserSysWait某银行Power在高峰时段(::)CPU利用率•CPU,GBMemory•NoVirtualization,DedicatedIO,OracleRACwithInfinibandIBMPowerSystems传统的服务器利用、扩展及整合结果是:所支付的硬件、软件、维护、机房空间和能源都被浪费了所支付的所利用的典型的UNIX或x服务器运行一个单独的应用大约利用率‡为未来增长预留配置(未利用)‡为业务高峰预留配置(未利用)‡即使系统在工作但等待IO和内存访问(未利用)PartitionedApplicationServerwitheightfixedpartitions物理分区整合ƒ在单个系统上的单一工作负载–平均利用率:ƒ个独立的工作负载在一个物理分区的系统上–平均利用率仍为ƒ在一个物理系统上倍的硬件资源结果是:所支付的倍的硬件和软件被浪费了由于没有资源共享利用率仍然相同SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)结果是:所支付的倍的硬件、软件、维护、机房空间和能源被浪费了向外扩展•在单个系统上的单一工作负载假定平均利用率为高峰期为倍高峰是随机的•八个独立的工作负载位于八个同样的系统上相同的假设IBMPowerSystemsAPPAPPAPPAPPPRDB#PRDB#TestTestQADB#QADB#QAQAPowerVM虚拟化技术提高系统利用率不使用分区或仅使用DLPAR技术使用微分区和共享处理器池技术IBMPowerSystems同样场景采用PowerVM虚拟化技术整合(共享CPU资源)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)SingleApplicationServer(cores)toSystemsConsolidation(cores)ƒ八个独立的工作负载位于八个同样core的系统上–平均利用率:ƒ八个独立工作负载位于一个core系统上–平均利用率:ƒ核减少到核(>)toSystemsConsolidation(cores)ƒ个独立的工作负载位于个同样core的系统上–平均利用率:ƒ个独立工作负载位于一个core系统上–平均利用率:ƒ核减少到核(>)ƒ个独立的工作负载位于个core同样的系统上–平均利用率:ƒ个独立工作负载位于一个core系统–平均利用率:ƒ核减少到核(>)toSystemsConsolidation(cores)#ofworkloads#ofCPUstototoƒ平均利用率随着工作负载的增加而不断提高()>()>()ƒ随着工作负载整合数量的增加成本节约更加显著IBMPowerSystemsCRMDBFailoverQABWDBFailoverDevERPDBFailoverTestCRMDBBWDBERPDBProductionSystemsHighAvailabilityExampleSharedDiskQATestDevCRMDBFailoverBWDBFailoverERPDBFailoverCRMHAERPHABWHAQADevTestCRMDBBWDBERPDBCRMDBFailoverBWDBFailoverERPDBFailoverIBMPowerSystemsƒSupportforPOWERandPOWERsystemsHMCVRistheminimumlevelforPOWERsupportPOWERsystemsshouldhaveGA(SF)FirmwareƒHMCshippedwithPOWERwillhavenewMachineTypeCxandCRx(desktoprackmounted)willbepreloadedwithHMCVersioncodeƒAdditionalcapabilitieswithPOWERServersWillsupportPOWERspecifics,suchasLPARMigrationƒHMCVersionnewReleaseofHMCcodeSupportedonexistingmodelsofHMCƒUpgradeofexistingHMCOrderFC,andupdateMCRSAmaintenancesupportHMCusedtomanageanyPOWERprocessorbasedserver,mustbeaCRorlatermodelrackmountHMCorCorlaterdesksideHMCPOWERHMCEnhancementsIBMPowerSystems动态分区ƒ所有PowerSystem都具有的能力HMCAIXLLinuxHypervisorPart#ProductionPart#Part#Part#LegacyAppsTestDevFilePrintAIXLAIXLMoveresourcesbetweenlivepartitionsIBMPowerSystems微分区技术分区类型–微分区:最大个*–动态分区:最大个*–以上两种的组合通过HMC配置虚拟CPU的个数–最小最大CPU处理能力–最小颗CPU–最小颗CPU增加受限和非受限分区MicropartitionsPoolofCPUsLinuxiOSVR**AIXLVAIXLVLinuxEntitledcapacityHypervisorMinMax*onpandp**onp,p,andpAIXLVAIXLVDynamicLPARsWholeProcessors微分区技术允许把一颗CPU分成份,每一个分区最少是颗CPU,帮助整合UNIX和Linux应用Note:MicropartitionsareoptionalIBMPowerSystems共享CPU共享CPU概念z每个分区根据分配的CPU处理能力获得物理CPU的处理时间并且根据虚拟处理器个数分成几部分z个分区分配的最小CPU处理能力是个CPUz分区的CPU处理能力叫做entitlement,orentitledcapacity定义为能够获得多少物理CPU处理时间IBMPowerSystemsSharedprocessorpool处理器术语SharedprocessorpartitionSMTOffSharedprocessorpartitionSMTOnDedicatedprocessorpartitionSMTOffDeconfiguredInactive(CUoD)DedicatedSharedVirtualLogical(SMT)InstalledphysicalprocessorsEntitledcapacityIBMPowerSystems微分区ƒ微分区支持多个分区共享一个物理CPUƒ一个物理CPU最多支持个分区ƒ一台机器最多支持同时启动个分区ƒ分区资源定义–每个资源有最小值、最大值和理想值–CPU的处理能力–虚拟处理器个数–受限或者非受限•不受限分区分配优先级–专用的内存•最小MB并且可以每MB增加–物理的或者虚拟的IO资源CPUCPUCPUCPULPARLPARLPARLPARLPARLPARIBMPowerSystems„LPARsaredefinedtobededicatedorsharedƒDedicatedpartitionsusewholenumberofCPUsƒSharedpartit

用户评价(0)

关闭

新课改视野下建构高中语文教学实验成果报告(32KB)

抱歉,积分不足下载失败,请稍后再试!

提示

试读已结束,如需要继续阅读或者下载,敬请购买!

文档小程序码

使用微信“扫一扫”扫码寻找文档

1

打开微信

2

扫描小程序码

3

发布寻找信息

4

等待寻找结果

我知道了
评分:

/16

IBM_PowerVM虚拟化介绍及案例经验

仅供在线阅读

VIP

在线
客服

免费
邮箱

爱问共享资料服务号

扫描关注领取更多福利