BMC_BSM解决方案介绍.ppt

基于配置管理数据库的服务模型的服务影响管理 有效的事件处理系统 支持平台广泛、抓取数据详尽 支持应用数据的抓取 请问您今天有几片叶子？ 每片叶子是哪产的？ 哪片叶子是您今天最不满意的？ 您有树干吗？ 您的树干有根吗？ 监控管理平台设计思路 网络管理平台设计思路 基于SNMP的监控体系，区别于系统监控的管理模式 是一个独立的管理内容 内容包括：性能管理、故障（告警）管理、配置管理及报表管理 需要与系统管理平台及其他管理平台的整合，进行集中管理 考虑在下一期项目中，与服务影响分析管理的整合 系统管理平台设计思路 采用Agent及Agentless方式的混合模式 支持平台广泛 监控内容详细 资源占用小 能够进行客户化 运行维护简单 监控管理平台设计目标 减少故障诊断时间 减少故障处理时间 减少故障对业务的影响 提高预警技术，减少故障 * * * * * * * * Most performance monitoring solutions have a threshold/event based approach that was originally designed to monitor availability. This approach uses a threshold to divide performance into good (below threshold) and bad (above threshold). This approach has significant drawbacks: Requires establishing hundreds if not thousands of thresholds and rules that require a lot of maintenance as the environment evolves and grows. Thresholds generate too many events/alarms when they get exceeded. So the tendency is to set them very high. But, by doing so you miss the visibility into the behavior below the threshold. ProactiveNet self-learns Dynamic Baselines The ProactiveNet Analytics approach is to self learn “Dynamic Baselines” which are a “band of normal operation” for every metric we collect. Within a short period of time (a matter of days to a couple of weeks), ProactiveNet Analytics knows what is ‘normal’ behavior and recognizes, automatically, in real-time, when any of the monitored metrics becomes abnormal. This approach dramatically reduces the cost of setting up and operating the system versus in a solution where you have to manually set and maintain thresholds and rules. With ProactiveNet’s Dynamic Baselines and Correlation, you would recognize the beginning of the problem when performance is beginning to degrade and you would be able to reduce the amount of time needed to isolate, diagnose and resolve the problem before service delivery or customer are impacted. This is the fundamental difference in the approach used by ProactiveNet versus threshold/event based solution