天津移动业务支撑应急系统设计与实现.docxVIP

最新 精品 Word 欢迎下载 可修改 摘 要 目前中国移动集团天津公司NGCRM/BOSS系统的业务连续性保障体系有三种模式，一种是多节点负荷分担方式，该方式主要用于系统接入层和业务逻辑层，有效地降低了个别节点故障对业务的影响程度；一种是容灾模式，由于多年未升级，系统资源与生产中心已不匹配，在发生突发事件时，容灾系统不能在特定的时间要求内全部或局部恢复关键业务功能；一种是双机备份共享存储〔以下简称本地HA〕方式，该方式主要用于系统核心层。对于系统核心层采用的本地HA模式来保障业务连续性，存在如下风险： 1) 由于核心系统IO量较大，如发生系统单节点宕机等严重故障可能会造成由于IO未及时写入磁盘而产生的文件系统错误，导致备机启动失败。 2) 人为因素数据库逻辑错误或者存储故障造成的数据损坏从而引起业务中断，本地HA将无法解决。NGCRM/BOSS系统全部业务要求7×24小时运行，存储阵列的使用强度大大增加，没有时间对存储系统进行定期维修和保养。因此，当使用一段时间后，存储系统的部件连续或同时出现故障的可能性增加。此外，随着存储系统的功能和性能越来越强，存储系统内部的控制软件也日趋复杂，就像一个操作系统，其本身也会出现故障或漏洞。局部省公司也曾经发生过由于存储故障造成业务系统长时间停机数据丧失的重大故障。 3) 在系统割接平台软硬件维护或应用版本升级等情况下，本地HA都将可能无法满足业务连续性要求。 4) 生产机房发生火灾泡水等情况下，多节点负载分担和本地HA模式都不能保障业务连续性。 本文将从应急系统的系统架构建设实现系统测试各方面对于上述风险及问题进行研究并逐一解决。 ** 业务支撑系统　应急系统　运营商 ABSTRACT At present the Tianjin NGCRM/BOSS business continuity security system has three modes, one is a multinode load balancing mode, this mode is mainly used for system access layer and business logic, effectively reducing the individual node failuresthe degree of influence of the business; a disaster recovery mode, due to years of not upgraded, the system resources and production center does not match, not within a specific time requirements in whole or in part, to restore critical business functions in the event of an emergency, disaster recovery system; a double backup shared storage (hereinafter referred to as the local HA) mode, which is mainly used for the core of the system layer. The local HA mode for the system core layer to protect business continuity, the following risks: 1) due to the large amount of core system IO, such as the occurrence of a serious failure of the system singlenode downtime may cause IO is not written to disk errors, leading to the backup machine failed to start. 2) data corruption caused by human factors, database logic error or storage failure causing business interruption, local HA will not resolve. All of NGCRM/BOSS system requirements 7 × 24 hours to run, greatly increase the intensity of use of the storage array, do no