基于CDVRS的虚拟机资源调度策略研究

The research of virtual machine resources scheduling based on CDVRS in cloud computing

云计算将各种处理器、存储以及网络等物理资源虚拟化为虚拟机,为用户分配相应的虚拟机以及将虚拟机调度到物理资源上是云计算中一个重要问题.提出一种基于分类挖掘的虚拟机资源调度模型及算法CDVRS(virtual machine resources scheduling based on classification data mining)解决该问题,采集用户访问虚拟机及虚拟机映射物理资源的历史信息,采用改进的分类挖掘算法对其进行挖掘,得到指导虚拟机资源调度的分类规则和模式,在此基础上实施虚拟机资源调度.对模型和算法进行仿真,实验结果表明采用CDVRS算法相比GA(genetic algorithm)、PSO(particle swarm optimization)等算法在资源利用率上有较大的改善,能有效提高云计算中虚拟机资源调度的效率.

The physical resources were usually abstracted to virtual resources in cloud computing, such as computing resources, storage resources and network resources etc. How to schedule these virtual machines and all kinds of physical resources in clouds based on users＇ service requests is important, and it can affect the efficiency of cloud computing platform. A virtual machine resources scheduling algorithm based on classification data mining （CDVRS） was proposed in this paper. At first, we gathered and analyzed the historical data of users＇ accessing to virtual machines and virtual machines map to physical resources, and then got the classification rules and patterns by application of the improved classification mining algorithm. Finally, we scheduled new arrival jobs to virtual machines in cloud computing based on these classification rules. Experiments showed that CDVRS algorithm has better performance in virtual machines and physical resources utilization compared with other algorithms, and it could improve the efficiency of resources in cloud computing environment.