Abstract: Virtual machine management is a crucial mission of the cloud datacenters.It identifies?the?mapping between
virtual machines and physical machines，which has?important?implications?for clustering?performance，response?time and service quality，etc.It is necessary to adopt effective strategies to make sure it can be completed automatically and goals such as load balancing，virtual machine migration and power consumption are achieved at the same time.This paper proposes an improved artificial fish swarm algorithm which employs a new behaviors selection and applies it to solve the virtual machine management problem.The?simulation?calculation results，compared?with?other common algorithms，show that it can obtain high quality solution with much less time consumption without accuracy losing.

Key words: virtual machine management, load balancing, power consumption, artificial fish-school swarm algorithm, Behaviors selection, combinatorial optimization

摘要： 虚拟机分配是云数据中心的一个重要任务，它实现物理机和虚拟机的映射，这对整个云数据中心中计算集群的性能，响应时间和服务质量有重要影响。需要采用一定的虚拟机分配策略来保障在同一集群中自动地完成虚拟机分配，以便达到物理机集群的负载均衡，虚拟机迁移次数最小并且节能环保等目的。提出了一种添加新型行为策略的人工鱼群算法，并将该算法成功应用于虚拟机分配问题的求解。与多种虚拟机分配算法的对比结果表明，算法能更快获得符合云数据中心多个需求的分配方案，对于其他实时性要求较高的组合优化问题同样具有应用价值。

关键词: 虚拟机调度, 负载均衡, 节能环保, 人工鱼群算法, 行为策略, 组合优化