01WXSHighlyScalableApps中文新编.pptVIP

问题? Stipe as the RAID5 Next few slides we’ll be going back and forth. If you don’t understand, please ask questions * Each cache is local to it’s own application server. They do not communicate with each other. For example, Application server A has no knowledge of the objects stored in cache on application server B. * Benefit: The memory is as large as you want it to be (as large as you make the grid). Some IBM customers have ALL objects stored in the grid, so requests from the application server never go directly to the database. In this case, the DB is only used for disaster recovery (in case the grid completely fails). * In this case, the client is an application within one of the application servers. * If the WXS grid is a cache of the DB, and WXS clients are a cache of the grid, then WXS clients are a cache of a cache * Remember, there is typically 1 container per WXS server instance (JVM). However, it is possible to create multiple containers per server instance. * Note partitions start with #0. Note, the record icons disappear for simplification. The data still exists on each shard. * This slide transitions into the catalog service * This slide transitions into stopping a container (or a failed container) and then adding a new container to the grid. * Note, some layers may be hidden here. View in slide mode to see the correctr representations of the shards. 议题 性能扩展的挑战和困惑 eXtreme Scale (WXS)基本概念 eXtreme Scale (WXS)架构 WXS 对象操作的内部原理和机制 WXS用户案例 Scalable Architecture可扩展的体系架构 WebSphere eXtreme Scale 先进的应用基础架构 议题 性能扩展的挑战和困惑 eXtreme Scale (WXS)基本概念 eXtreme Scale (WXS)架构 WXS 对象操作的内部原理和机制 WXS用户案例 * DB 集群服务器访问DB的样例拓扑结构 每个 代表一个应用服务器实例(或JVM实例), 比如 WebSphere Application Server应用服务器实例。 并由它们组成一个集群 每个 代表一个从应用服务器到数据库的访问请求, 比如：读，写，更新等 Application Servers 存在的问题: 在这样一个普遍适用的拓扑环境中，数据库由于锁冲突、磁盘I/O、网络请求等等原因可能会成为系统的性能瓶颈 ! * Application Servers DB 使用了本地缓存的同样拓扑架构 每个 代表一个本地的缓存，比如使用Java中的HashMap，这样每个服务器实例的应用中将会有一个它自身使用的本地缓存。 每个 代表Java数