容灾技术培训核心技术及概念解释.ppt
《容灾技术培训核心技术及概念解释.ppt》由会员分享,可在线阅读,更多相关《容灾技术培训核心技术及概念解释.ppt(95页珍藏版)》请在三一办公上搜索。
1、容灾技术研讨,核心技术及概念解释,客户需求分析容灾的目的,业务不停顿业务连续性通常不存在不停顿业务的容灾手段群集(应用层面)数据复制(存储层面)数据不丢失数据完整性理论上不存在不丢失数据的容灾手段备份(生产系统某个时间点的静态数据拷贝)复制(动态跟踪生产系统的数据变化)通常用户的需求覆盖上述二个方面,业界对容灾实施的标准,Recovery Point Objective(RPO)The point to which data must be restoredAmount of data loss acceptableRecovery Time Objective(RTO)The time by
2、 which data must be restoredAmount of time it takes to come back online,成功容灾的基本技术要素,应用程序应用程序本身必须是Fault Torrent的群集技术群集软件必须能够正确地切换,并支持底层的存储复制技术存储技术保证数据的连续性保证数据的一致性,在容灾中所用到的技术,Backup,Replication,Mins,Days,Hrs,Secs,Wks,Days,Mins,Hrs,Wks,Secs,Data Loss(Recovery Point Objective),Downtime(Recovery Time Obj
3、ective),Restore from Tape,Clustering+Replication,Restore from Disk,Vaulting,Protecting data from the desktop to the data center,Storing backup tape copies offsite for later recovery,Copying data to another location for instant access at a secondary site,如何满足用户需求不丢失数据,Backup,Replication,Vaulting,如何满足
4、用户需求业务不停顿,Recovery of systems and data with significant downtime,Automatically recover systems and data with minimal downtime,Restore from Disk,Clustering+Replication,Restore from Tape,Mins,Days,Hrs,Secs,Wks,Downtime(Recovery Time Objective),Recovery of systems and applications with no downtime,容灾方案
5、备份/恢复,备份/异地磁带保存VERITAS NetBackupVERITAS NetBackup Vault Option,恢复Tape and Disk RestoreReplicate Backup CatalogVERITAS Bare Metal Restore,容灾解决方案群集,本地群集/城域群集/远程应用切换,Remote Data ReplicaVERITAS Volume ManagerVERITAS Volume Replicator,Application AvailabilityVERITAS Cluster Server,如何提高用户业务系统的可靠性,remote c
6、lustering,remote replication,local clustering,campus mirroring,backup,availability,容灾技术研讨,方案选择/部署,关注系统潜在风险,一个好的解决方案应能考虑到各种风险情况:硬件失效 应用程序错误人为错误 病毒干扰IT系统的升级、割接 新技术手段的应用自然灾害“各种无法预见的事件”,提高系统可用性的技术,高可用性,投资,可靠的备份方案,磁盘和卷管理,局域环境设施,管理客户端,管理服务和应用程序,故障冗余,数据复制,灾难恢复,优良的系统管理经验,网络(LAN和SAN),Secondary Site,Applicati
7、on,VERITAS Volume Manager,Primary Site,VERITAS Volume Manager在城域光纤网上进行阵列间的同步镜像,Description:利用扩展的SAN网络将数据同步复制到灾备节点 Benefits:节省投资通过SAN网络自然地镜像数据(无需任何特殊附加设备)可以任意选择磁盘阵列类型系统保护特性基本没有数据丢失可管理性在所有OS平台上具有统一的管理界面,对用户及上层应用完全透明可以随时监控灾备节点数据完好性,技术实现机制,Volume 1,PlexVolume1-01,PlexVolume1-02,Mirror,Application/File S
8、ystem,OS Devices,I/O,I/O,I/O Control,I/O PATH,VERITASVolume Manager,潜在的问题,存储性能传输距离如何实施意外情况带来的影响.But在任何情况下,都等同于磁盘阵列做同步复制,Secondary Site,Application,VERITAS Volume Manager,Primary Site,VERITAS Cluster Server城域范围内的群集系统,VERITAS Volume Manager,Application,Cluster Server,Cluster Server,Description:将局域网内的群
9、集系统扩展到城域范围 Benefits:节省投资直接利用现有的群集系统系统保护特性秒/毫秒级的数据丢失率,大部分情况下没有数据丢失分钟级的系统停机时间(通常在1分钟30秒内完成灾备节点应用接管)可管理性无需对应用系统作任何改动,切换方式和本地群集没有区别可以制定灵活的策略实现多级群集,Primary Site,Secondary Site,VERITAS Volume Replicator通过IP网络实现数据复制,Description:允许数据通过IP网络,以同步/异步的方式进行复制Benefits节约成本无磁盘阵列种类限制 节省存储空间 系统保护特性Uncompromised Data I
10、ntegrity in any modeFull database support(all modes)Protection against network outages可管理性Same management across operating systemsAbility to utilize primary and secondary sitesGet secondary site up and running quickly and easily,VVR 技术机制,Replicated Volume Group(RVG)Rlink(s)Object abstraction for Oth
11、er NodeData Volume(s)Group of Related VolumesAvailable for applicationsReplicator Log(SRL)Not available for applicationsCircular buffer to store updatesReplicated Data Set(RDS)Groups Primary RVG and Secondary RVGs,RLink,IP Network,Volume Replicator:同步模式,com,ack,64kb,Exact mirror copy at secondary lo
12、cationApplication Performance impact is equal to network latency,64kb,com,IP Network,56kb,Volume Replicator:异步模式,com,ack,64kb,com,No application performance impactMinimal Data Loss typically millisecondsProtect against network outagesUtilize Primary and Secondary SitesFail over and fail back easily
13、between sites,com,80kb,56kb,Volume Manager,1 m 10 m 1 km 10 km 100 km 1000 km 10,000 km,Volume Replicator,如何让容灾方案适合更复杂的环境,Choose Volume Manager:Mirroring over Fibre ChannelMirroring within a MANMirroring less then 100kmMirroring synchronously,Choose Volume Replicator:Replicating over IP NetworkNo ex
14、tended SAN InfrastructureReplicating over WANThe most robust and flexible technology for replicating greater than 100kmReplicating synchronously or asynchronously,整合用户应用环境,Utilization:,Traditional Approach,优化容灾资源配置,Volume Manager,1 m 10 m 1 km 10 km 100 km 1000 km 10,000 km,Volume Replicator,如何让容灾方案
15、适合更复杂的环境,Choose Volume Manager:Mirroring over Fibre ChannelMirroring within a MANMirroring less then 100kmMirroring synchronously,Choose Volume Replicator:Replicating over IP NetworkNo extended SAN InfrastructureReplicating over WANThe most robust and flexible technology for replicating greater than
16、 100kmReplicating synchronously or asynchronously,容灾技术研讨,各种容灾方式的技术细节,流行的容灾方式,容灾其实就是数据的复制(至少狭义的观点是这样)按容灾层面分:应用级容灾事实上不存在应用级容灾,仅仅是数据库厂商的一些概念存储级容灾基于底层数据复制(更准确的说法是基于data block的复制)按实现的技术手段分:基于主机:VERITAS基于SAN网络设备基于磁盘阵列按数据复制的方式:同步异步周期但是,数据复制技术的大致领域,首先让我们探讨一个最基本的问题,数据的连续性和实时性(Current and Consistent),Data Vol
17、ume 1,Data Volume 2,Data Volume 3,Data Volume 1,Data Volume 2,Data Volume 3,Primary host,Secondary host,数据的连续性和实时性(Consistent,Not Current),Data Volume 1,Data Volume 2,Data Volume 3,Data Volume 1,Data Volume 2,Data Volume 3,Primary host,Secondary host,数据的连续性和实时性(Not Consistent,Not Current),Data Volum
18、e 1,Data Volume 2,Data Volume 3,Data Volume 1,Data Volume 2,Data Volume 3,Primary host,Secondary host,数据复制的技术内核,Bitmap of regions on a primary volume that differ from the secondary volumeNo information on write orderRecovery involves copying all regions marked in the DCM to the secondary,Marks a pos
19、ition in the primary SRL for the start of the copyWhen the copy is complete,mark the end on the SRLReplay all changes written to the primary RVG during the copy to the secondary RVG from the SRL,不同的技术内核数据一致性探讨,BITMAP方式:实时性好无法保证数据一致性,利用数据冗余解决带时间戳的LOG方式:数据一致性好很难在保证性能的情况下实现实时的数据复制,Synchronous Replicati
20、on,Primary RVG,Secondary RVG,VVR的同步工作模式,Write goes to primary SRLData is sent to VVR memory on secondary RVGDestage to local disk on primary RVGSecondary RVG acknowledges receipt(network acknowledgement)Acknowledge application Secondary RVG acknowledges data written to local disk(data acknowledgemen
21、t),VVR Layer,VVR Layer,Primary RVG,SRL,Application Layer,RLINK,SRL,RLINK,VVR Memory,同步复制情况下的时延考虑,Total latency in synchronous replication involves 4 distinct areasLocal write to SRLTransmission delay to secondary RVGReceive,process,and send network acknowledgementTransmission delay to primary RVGIte
22、ms 2 and 4 usually take the most time,VVR Layer,VVR Layer,Primary RVG,1,Application Layer,RLINK,SRL,RLINK,VVR Memory,2,3,4,Asynchronous Replication,Primary RVG,Secondary RVG,异步模式下的数据流,Write goes to primary SRLAcknowledge application Data is sent to VVR memory on secondary RVGDestage to local disk on
23、 primary RVGSecondary RVG acknowledges receipt(network acknowledgement)Secondary RVG acknowledges data written to local disk(data acknowledgement),VVR Layer,VVR Layer,Primary RVG,SRL,Application Layer,RLINK,SRL,RLINK,VVR Memory,数据复制方式,容灾系统技术架构的选择,技术实现手段,基于主机的数据复制,数据卷镜像,同步 异步,基于数据库的复制,基于存储硬件数据复制,基于存储
24、网络的数据复制,应用,主服务器,备服务器,复制器,基于存储网络的数据复制,基于存储网络的虚拟化支持异构主机和存储环境集中管理可读写的数据快照脱机操作用来做备份处理在生产和备用中心之间同步数据本地和城域范围的数据镜像未来支持远程数据复制,场地 A,场地B,基于存储网络的数据复制,场地 A,场地 B-10s Km同步数据复制,IP,基于存储硬件的数据复制,应用,网络,主服务器,备服务器,复制器,复制器,优势操作系统支持弱势没有存储平台选择空间锁定在原供应商上存储成本高居不下传输距离有限在IP网上传输需要特殊设备同步模式现场测试表明卷管理软件的数据复制性能要高于硬件数据复制异步模式担心数据完整性“异
- 配套讲稿:
如PPT文件的首页显示word图标,表示该PPT已包含配套word讲稿。双击word图标可打开word文档。
- 特殊限制:
部分文档作品中含有的国旗、国徽等图片,仅作为作品整体效果示例展示,禁止商用。设计者仅对作品中独创性部分享有著作权。
- 关 键 词:
- 技术培训 核心技术 概念 解释
链接地址:https://www.31ppt.com/p-2233240.html