一种面向云环境虚拟机部署的粒子群优化策略

Particle Swarm Optimization Strategy for VM Placement in Cloud

随着数据中心规模的扩大,提升资源利用率的问题日益凸显.如何合理部署虚拟机到数据中心的物理机上是提升数据中心资源利用率的关键问题之一.首先提出相应的资源利用率和响应时间模型,并提出一种结合变异算子的多目标粒子群优化算法(Multi-Objective Particle Swarm Optimization with Mutation Operator,MOPSOM)用于解决虚拟机部署问题.该算法以提高资源利用率和减少响应时间为优化目标,采用线性递减惯性权重寻找最优的虚拟机分配方案.仿真实验结果表明,MOPSOM能够在减小响应时间下提高资源利用率,同时在负载均衡度和能耗之间达到了更好的平衡.

With the expansion of the data center,the problem of increasing the utilization of resources is becoming increasingly prominent.How to deploy the virtual machine to the data center is one of the key issues to improve the resource utilization of the data center.Firstly,the corresponding resource utilization and response time model are proposed,and a multi-objective particle swarm optimization algorithm with mutation operator（ MOPSOM） is proposed to solve the problem of virtual machine deployment. In order to improve the efficiency of resource utilization and reduce the response time,the algorithm uses linear decreasing inertia weight to find the optimal virtual machine allocation scheme. The simulation results show that MOPSOM can improve the efficiency of resource utilization in reducing response time,and achieve a better balance between load balance and energy consumption.