《Exchange_Server白手起家系列_系统SLA和监控流程.pptx》由会员分享，可在线阅读，更多相关《Exchange_Server白手起家系列_系统SLA和监控流程.pptx（37页珍藏版）》请在taowenge.com淘文阁网|工程机械CAD图纸|机械工程制图|CAD装配图下载|SolidWorks_CaTia_CAD_UG_PROE_设计图分享下载上搜索。
1、Exchange Server白手起家系列白手起家系列系统系统SLA和监控流程和监控流程喻喻 勇勇,PMP/MCSEPMP/MCSE微软最有价值专家，特约讲师微软最有价值专家，特约讲师微软最有价值专家，特约讲师微软最有价值专家，特约讲师系列课程内容系列课程内容uu万事开头难万事开头难 uu进行企业进行企业ITIT架构评估架构评估 uu做一个做一个ExchangeExchange管理员必须掌握的技术管理员必须掌握的技术-软件篇软件篇 uu做一个做一个ExchangeExchange管理员必须掌握的技术管理员必须掌握的技术-硬件篇硬件篇 uu邮件系统设计和选型的原则邮件系统设计和选型的原则 uu如
2、何执行压力测试？如何执行压力测试？uu软件安装和服务器管理的规范化软件安装和服务器管理的规范化uu用户和系统迁移过程中的注意事项用户和系统迁移过程中的注意事项uu大规模的用户帐号管理、组管理中采用的常见策略大规模的用户帐号管理、组管理中采用的常见策略uu客户端部署和管理客户端部署和管理uu邮件系统日常维护的要点邮件系统日常维护的要点uu系统系统SLASLA和监控流程和监控流程uu邮件安全、反病毒和反垃圾邮件邮件安全、反病毒和反垃圾邮件uu存储备份策略存储备份策略uu你的系统可靠性有几个你的系统可靠性有几个9 9？uuDownDown机了怎么办？机了怎么办？uu常见问题汇总和答疑常见问题汇总和答
3、疑今日议程今日议程uuSLA是反映一项是反映一项IT服务的综合指标，不仅仅代服务的综合指标，不仅仅代表系统的在线时间，更反映了用户对系统的认表系统的在线时间，更反映了用户对系统的认可程度。本次课程主要讨论构成可程度。本次课程主要讨论构成SLA的组成部的组成部分，以及如何通过有效的工作来提升分，以及如何通过有效的工作来提升SLA和用和用户的满意程度。户的满意程度。uuSLA的概念和深入解析的概念和深入解析uu关键业务系统监控的流程关键业务系统监控的流程uuExchange主要的性能指标主要的性能指标uu性能优化建议性能优化建议SLA的概念和深入解析的概念和深入解析uuSLA：Service Le
4、vel Agreement uu是在确定的成本下与客户达成的服务和系统使是在确定的成本下与客户达成的服务和系统使用情况的协议。用情况的协议。uu最简单的指标：系统服务在线时间最简单的指标：系统服务在线时间uu更进一步的衡量：更进一步的衡量：资源：软硬件的使用效率和投资回报资源：软硬件的使用效率和投资回报资源：软硬件的使用效率和投资回报资源：软硬件的使用效率和投资回报性能：用户的使用体验满意程度性能：用户的使用体验满意程度性能：用户的使用体验满意程度性能：用户的使用体验满意程度服务：各类问题和帮助请求的解决速度服务：各类问题和帮助请求的解决速度服务：各类问题和帮助请求的解决速度服务：各类问题和帮
5、助请求的解决速度服务级别协议服务级别协议(SLA)的结构的结构uuService Based-One Service Level Agreement for Service Based-One Service Level Agreement for a specific service for all customers of that a specific service for all customers of that serviceserviceuuCustomer Based-One Service Level Agreement Customer Based-One Service
6、 Level Agreement for a specific customer for all servicesfor a specific customer for all servicesuuMulti-LevelMulti-LevelCorporate Level-covering all generic Service Level Management Corporate Level-covering all generic Service Level Management(SLM)issues that would be appropriate for all customers(
7、SLM)issues that would be appropriate for all customersCustomer Level-covering all Service Level Management(SLM)Customer Level-covering all Service Level Management(SLM)issues relevant to a specific customer group for all services usedissues relevant to a specific customer group for all services used
8、Service Level-covering all Service Level Management(SLM)Service Level-covering all Service Level Management(SLM)issues relevant to a specific service for a specific customer groupissues relevant to a specific service for a specific customer group如何制定如何制定SLAuu在项目实施阶段的前期了解用户对在项目实施阶段的前期了解用户对SLA的需求的需求uu
9、在实施过程中根据实际的软硬件投入情况进行在实施过程中根据实际的软硬件投入情况进行动态的更新，并及时与客户动态的更新，并及时与客户(用户用户)进行沟通进行沟通uu如果项目涉及到第三方供应商，应在制定如果项目涉及到第三方供应商，应在制定SLA之前取得第三方的认可和确认之前取得第三方的认可和确认如何贯彻如何贯彻SLAuu服务器端服务器端服务器端服务器端服务器配置和部署的规范统一服务器配置和部署的规范统一服务器配置和部署的规范统一服务器配置和部署的规范统一运营过程中遵循严格的流程运营过程中遵循严格的流程运营过程中遵循严格的流程运营过程中遵循严格的流程变更管理变更管理变更管理变更管理uu最终用户最终用户
10、最终用户最终用户培训培训培训培训HelpdeskHelpdesk丰富客户端的功能，增强用户体验丰富客户端的功能，增强用户体验丰富客户端的功能，增强用户体验丰富客户端的功能，增强用户体验uu整体协调整体协调整体协调整体协调ITIT支持部门内部的横向沟通和协调支持部门内部的横向沟通和协调支持部门内部的横向沟通和协调支持部门内部的横向沟通和协调管理与第三方供应商的关系管理与第三方供应商的关系管理与第三方供应商的关系管理与第三方供应商的关系uu及时保持与客户及时保持与客户及时保持与客户及时保持与客户(IT(IT主管部门，主管部门，主管部门，主管部门，CIOCIO等等等等)的沟通的沟通的沟通的沟通如何衡
11、量如何衡量SLAuu服务器在线时间服务器在线时间uu客户满意度客户满意度服务器在线时间服务器在线时间故障发生检测到问题检测时间诊断响应时间修复恢复服务问题解决下一次故障修复时间修复时间恢复时间Mean Time to Repair(MTTR)“总的停机时间Time Between System Incidents(MTBSI)Time Between Failures(MTBF)客户满意度客户满意度uu客观的设定用户的期望值客观的设定用户的期望值uu建立良好的建立良好的Helpdesk流程流程uu如果是跨国企业，如果是跨国企业，Helpdesk流程的一致性非流程的一致性非常重要常重要性能性能可
12、扩展性可扩展性可用性可用性关键业务系统监控的流程关键业务系统监控的流程uu每日检查的内容每日检查的内容备份情况备份情况备份情况备份情况邮件队列邮件队列邮件队列邮件队列系统性能系统性能系统性能系统性能磁盘空间磁盘空间磁盘空间磁盘空间硬件情况硬件情况硬件情况硬件情况昨天或昨晚的重要变更昨天或昨晚的重要变更昨天或昨晚的重要变更昨天或昨晚的重要变更uu每日检查报表的分发每日检查报表的分发建议使用建议使用建议使用建议使用SharePointSharePoint网站保存每日检查的报表网站保存每日检查的报表网站保存每日检查的报表网站保存每日检查的报表学会使用性能计数器来给学会使用性能计数器来给Exchang
13、e把脉把脉uudemo性能优化建议性能优化建议uuSmall design changes can reap big benefitsuuDo not be overly aggressive when tuning,especially when it involves registry changesuuAlways use the Microsoft Exchange Server Best Practices Analyzer Tool for validationStorage PerformanceuuThe biggest cause of Exchange performanc
14、e issuesThe biggest cause of Exchange performance issuesUsually too few spindles to meet the demandUsually too few spindles to meet the demanduuExchange is I/O hungryExchange is I/O hungryBack of the envelope:1 spindle for every 100 usersBack of the envelope:1 spindle for every 100 usersuuActions to
15、 takeActions to takeWork out your I/O profileWork out your I/O profileTell your storage vendor how many IOPS you will produceTell your storage vendor how many IOPS you will produceSet caching controllers to 100%write/4 Set caching controllers to 100%write/4 kilobyte(kilobyte(KB)cache page sizeKB)cac
16、he page sizeThink end-to-end storage(for example,Host Bus Adaptors)Think end-to-end storage(for example,Host Bus Adaptors)Use DiskPar to align tracks(performance boost)Use DiskPar to align tracks(performance boost)Use JetStress to validate your storage architectureUse JetStress to validate your stor
17、age architectureRead the paper,Read the paper,Optimizing Storage in Exchange 2003Optimizing Storage in Exchange 2003!Warning Signs!Average disk latency is greater than 20 millisecondsLong disk latency spikes are greater than 50 millisecondsFast Access to Active DirectoryuuIf Exchange is slow accessi
18、ng the Active DirectoryIf Exchange is slow accessing the Active Directory directory directory serviceservice,store performance suffers and mail delivery is slow,store performance suffers and mail delivery is slowuuDouble check that the Double check that the Directory Service Access(Directory Service
19、 Access(DSAccess)DSAccess)list only contains local domain controller or global catalog list only contains local domain controller or global catalog serversserversRemember that round robin is usedRemember that round robin is usedRound-trip times should be less than 10 msRound-trip times should be les
20、s than 10 msuuRule of thumbRule of thumbYou should have one global You should have one global catalogcatalog processor for every processor for every four Exchange processorsfour Exchange processorsuuVerify that the/3GB switch is being used on global Verify that the/3GB switch is being used on global
21、 catalogscatalogsIncreases JET cache from 512 megabytes(MB)to 1 Increases JET cache from 512 megabytes(MB)to 1 gigabyte(GB)gigabyte(GB)More of the Active Directory is cached in memoryMore of the Active Directory is cached in memoryCorrect Use of/3GB SwitchuuMailbox and public folder serversImplement
22、 when there is=1 GB RAMImplement when there is=1 GB RAMWindows Server 2003 Windows Server 2003 implement/USERVA=3030implement/USERVA=3030uuGlobal catalog serversImplement when there is=1 GB RAMImplement when there is=1 GB RAMuuFront-end and routing serversDo not implement regardless of memoryDo not
23、implement regardless of memoryInstead,tune memory for system cacheInstead,tune memory for system cache/3GB and/USERVA SwitchesUser ModeUser ModeKernel ModeKernel ModeUser ModeUser ModeUser ModeUser ModeKernel ModeKernel ModeUser ModeUser ModeHeapDeCommitFreeBlockThresholduuShould be set to 0 x40000
24、on all Exchange serversReduces virtual memory fragmentation by Reduces virtual memory fragmentation by coalescing small free memory regions into coalescing small free memory regions into larger blockslarger blocksEspecially important for large Exchange Especially important for large Exchange mailbox
25、 servers 9582 be gone!mailbox servers 9582 be gone!http:/ Enable Hyper-ThreadinguuSetting in BIOSuuSometimes defaults to OFFuuCan improve Exchange processor efficiency by 25%uuWorks with Windows Server 2003 and Windows 2000 ServerDeploy Exchange Server 2003 SP2uu30%reduction in I/O patternuuDrastic
26、reduction in log file replay timesFrom 30 seconds to less than 2 secondsFrom 30 seconds to less than 2 secondsuuSupport for 8-way hyper-threading serversuuSets MaxDSNSize to 10 MB by defaultuuSide:Detection of single-bit flip errorsHardware Device DriversuuNonPaged Pool and page NonPaged Pool and pa
27、ge table entry(PTE)efficiencytable entry(PTE)efficiencyuuDisable unused hardware devicesDisable unused hardware devicesEspecially consider network adapters and storage cardsEspecially consider network adapters and storage cardsuuUse identical hardware inside the serverUse identical hardware inside t
28、he serverOne driver for multiple devices One driver for multiple devices uuBe especially careful with graphics adaptersBe especially careful with graphics adaptersCan result in serious PTE reductionCan result in serious PTE reductionKeep drivers up-to-dateKeep drivers up-to-dateSwitch to VGA.SYS dri
29、verSwitch to VGA.SYS driveruuNeed more PTEs?Need more PTEs?For Exchange Server 2003,reduce/USERVA to 2970For Exchange Server 2003,reduce/USERVA to 2970For Exchange 2000 Server,use SystemPages registry For Exchange 2000 Server,use SystemPages registry parameter(steal from the Paged Pool)parameter(ste
30、al from the Paged Pool)!Red Flags!Free PTE count less than 5000Tune Antivirus ScannersuuGet latest advice from your antivirus vendor on optimum configuration for your serverNumber of ScanningThreadsNumber of ScanningThreadsMemory/disk file size thresholdMemory/disk file size thresholduuRemember to k
31、eep your antivirus binaries up-to-dateuuAvoids problems such asVirtual memory fragmentationVirtual memory fragmentationNon-paged pool errorsNon-paged pool errorsService failuresService failuresOnline Maintenance ScheduleuuSkew online maintenance for large mailbox servers so there is at least a 15 mi
32、nute gap between start timesuuSkew online maintenance and online backup schedule so that there is zero overlapBackup will kill maintenance threads if they collideBackup will kill maintenance threads if they collideuuAlways make sure that online maintenance gets a chance to runuuNever disable online
33、maintenanceLog Buffer TuninguuExchange Server 2003 defaults to 500 buffers(at Exchange Server 2003 defaults to 500 buffers(at 512 bytes)512 bytes)http:/ uuIf disk latencies are reasonable,but log stalls are If disk latencies are reasonable,but log stalls are high,increase log buffer value in Active
34、Directoryhigh,increase log buffer value in Active DirectoryThis is especially critical for deployments that use This is especially critical for deployments that use replicated storagereplicated storageuuUse a multiple of 64 and never more than 9000Use a multiple of 64 and never more than 9000Storage
35、 group greater than msExchESEParamLogBuffersMailbox Store CacheuuStore has a two hour cache for re-reading mailbox configuration settingsMailbox limits and thresholdsMailbox limits and thresholdsuuUse the Mailbox Cache Age Limit registry parameter to reduce wait timehttp:/ Use Dedicated Routing Serv
36、ersuuQueuing messages use paged pool memory and increase CPU utilizationuuEnsure that large mailbox servers can immediately offload messages to a local dedicated routing serveruuFor routing servers,split Simple Mail Transfer Protocol(SMTP)queue onto dedicated fast diskuuOnly implement CheckConnector
37、Restrictions registry parameter on routing serversReduce LinkState SizeuuAvoid creating or removing routing groupsUse REMONITOR inject to nullify old data if Use REMONITOR inject to nullify old data if necessary(Microsoft Product Support necessary(Microsoft Product Support Services tool)Services too
38、l)uuWatch out for oscillating connectionsUse SuppressStateChanges registry Use SuppressStateChanges registry parameter to prevent a server from parameter to prevent a server from broadcasting changesbroadcasting changeshttp:/ Avoid 7-bit ConversionsuuPairs of Exchange Server 2003 bridgeheads in a na
39、tive mode organization perform 8-bit binary MIME transfers to each other over Exchange Server Routing Group connectorsuuAll other combinations will force each message to go through 7-bit conversions resulting in 30%additional bytes on the wireImplement Message Size LimitsuuLarge messages can monopol
40、ize Exchange serversuuSet sensible limits in global settingsuuCan Exchange process messages of any size?uuOnly set SMTP level limits on incoming bridgehead serversuuImplement MaxDSNSize if you are not running Exchange Server 2003 SP1Check IP Stack ConfigurationuuWatch for typos on DNS and Windows In
41、ternet Name Service(WINS)addressesFall back to broadcastingFall back to broadcastinguuAlways ensure that primary DNS is availableThere is a 30%reduction in mail flow There is a 30%reduction in mail flow performance if secondary DNS is usedperformance if secondary DNS is usedUnderstand Client Add-Ins
42、Run ExBPA to Validate ChangesuuThe Microsoft Exchange Server Best Practices Analyzer(ExBPA)Tool is designed to analyze the configuration of your topology and servers as they are today and give you up-to-date recommendations for achieving maximum performance,scalability,and availabilityTechNet是什么是什么?
43、uu只需轻轻点击，答案就在您的指尖只需轻轻点击，答案就在您的指尖只需轻轻点击，答案就在您的指尖只需轻轻点击，答案就在您的指尖对于对于对于对于IT IT 专业人员来说，专业人员来说，专业人员来说，专业人员来说，TechNet TechNet 是一个知识的宝库，你可是一个知识的宝库，你可是一个知识的宝库，你可是一个知识的宝库，你可以找到关于如何规划，部署和管理微软产品的的技术资源以找到关于如何规划，部署和管理微软产品的的技术资源以找到关于如何规划，部署和管理微软产品的的技术资源以找到关于如何规划，部署和管理微软产品的的技术资源uu每月发放包含最新信息的每月发放包含最新信息的每月发放包含最新信息的每
44、月发放包含最新信息的 DVD DVD或者或者或者或者CDCD这是最权威的资源，可以帮助你评估、配置和维护微软产品。这是最权威的资源，可以帮助你评估、配置和维护微软产品。这是最权威的资源，可以帮助你评估、配置和维护微软产品。这是最权威的资源，可以帮助你评估、配置和维护微软产品。订阅订阅TechNetTechNet 网站网站uu两周发放一次的中文电子快报两周发放一次的中文电子快报两周发放一次的中文电子快报两周发放一次的中文电子快报安全更新安全更新安全更新安全更新,新的资源等等新的资源等等新的资源等等新的资源等等TechNet 中文电子快报中文电子快报uu有关最新微软产品介绍和技术的简报有关最新微软产品介绍和技术的简报有关最新微软产品介绍和技术的简报有关最新微软产品介绍和技术的简报uu上机试验上机试验上机试验上机试验,“,“如何操作如何操作如何操作如何操作”等信息等信息等信息等信息TechNet 活动活动和网站消息和网站消息uu用户群用户群用户群用户群uu可管理的新闻组可管理的新闻组可管理的新闻组可管理的新闻组中文社区中文社区我们从哪里可以了解到我们从哪里可以了解到 TechNet?