张小银+陈国胜
摘要:随着现代数据中心云计算规模日益增长,云计算平台的智能运维管理面临较大挑战,尤其在实时监控领域方面。本文首先对云计算监控技术进行了深入分析和研究,然后在开源云计算平台Hadoop环境下,将Ganglia和Nagois两种开源监控软件进行整合,并利用移动飞信来实现对云计算平台的实时监控。实验结果表明,该系统可对云计算平台内主机和服务以及运行环境的各项性能指标进行全方位监控,实现对故障的实时预警和报警,使得管理人员能准确定位、实时处理云平台异常情况,从而提高了云平台的服务质量,有较好的应用价值。
关键词:云计算;Hadoop;性能指标;监控系统
中图分类号:TP3 文献标志码:A文章编号:1672-1098(2015)02-0000-00
Abstract:With the growing scale of cloud computing in modern data centers, the intelligent operation and maintenance management faces a great challenge, especially in real-time monitoring. After a thorough analysis and research of cloud computing monitoring technologies, this paper integrates two open-source monitoring software Ganglia and Nagois in a Hadoop open-source cloud computing platform, and use a mobile message software FeiXin to achieve real-time monitoring of the cloud computing platform. Experimental results shows that the proposed system realizes an all-round monitoring of performance indicators for hosts and service of operating environment in cloud computing platform and a real-time warning of faults, which help management personnel accurately locate and real-timely process abnormal situations. Therefore the system improves the quality of service of cloud computing platform and has a good practical value.
Key words:Cloud computing; Hadoop; Performance indicators; Monitoring system
随着云计算技术的不断成熟发展,云计算平台的规模以及资源也不断增加。现代数据中心的运维管理面临着重大挑战,传统的管理方法和管理模式已经无法满足要求。
为提高云计算平台的可靠性,保证服务质量,有必要在云计算平台中引入监控机制[1-3] [10],以便能准确定位性能异常或故障的节点,及时做出恢复和调整;掌握整个系统的运行状况,分析系统瓶颈,为整个系统负载均衡提供数据支持,在系统出现异常时能起到预警的作用。
Ganglia是加州大学伯克利分校发起的一个开源监控项目,主要用来监控大规模分布式系统的性能[1] [2] [4] [10]。Nagios也是功能强大的开源监控系统,能监控所指定(本地和远程)的主机以及服务,可利用故障状态实现故障报警[1] [2] [3] [5] [10]。……