OpenStack云平台的监控系统算法设计与实现

Design and Implementation of Monitoring System Algorithm for OpenStack

随着虚拟化、云计算技术的迅速发展,传统的管理系统已经不能满足用户需求,因此,云计算平台的高效管理已逐渐成为社会关注的焦点。监控作为管理手段的重要举措之一,管理员能够通过监控收集系统运行的信息,并依据这些信息规划、调整云管理的行为,从而提供给用户更高的服务质量。文中部署了OpenStack云管理平台,并对传统的线性回归和马尔可夫链进行分析,总结出它们现有的弊端,并充分结合两种算法对云平台进行预测。由于OpenStack中的数据具有一定的规律性,在不同阶段采用不同的算法。实验结果表明,根据不同阶段采用不同的算法能够更加准确预测下一时刻的数据,从而减少数据在网络中的传输,减小了网络开销。

With the rapid development of virtualization and cloud computing technologies,traditional management systems can no longer meet users＇ needs.Therefore,efficient management of cloud computing platforms has gradually become the focus of social attention.Monitoring is one of the important measures of management methods.Administrators can collect information about the operation of the system through monitoring,and plan and adjust cloud management behavior based on this information,so as to provide users with higher service quality.In this paper,we deploy Open Stack,the cloud management platform,analyze the traditional linear regression and Markov chain,and sum up their existing drawbacks and predict the cloud platform by full combination of the two algorithms.Due to the regularity of data in Open Stack,we use different algorithm in different stages. Experiment shows that different algorithm in different stage can predict the data at the next moment more accurately,thus reducing the transmission of data in the network and the network overhead.