一种多目标蚁群优化的虚拟机放置算法

已有对数据中心虚拟机放置的研究大多为优化数据中心能源消耗和物理机资源浪费等,很少考虑数据中心网络流量的优化,有可能影响数据中心网络的扩展性.为了兼顾考虑物理机资源浪费和网络总流量两个方面,将虚拟机放置建模为多目标优化问题,同时优化2个目标:最小化物理机资源浪费以提高数据中心物理机使用效率;最小化网络总流量以改善数据中心网络的扩展性.设计了一种基于多目标蚁群优化的虚拟机放置算法来求解该问题.仿真实验结果表明,该算法与首次适合递减算法相比降低了物理机资源浪费和网络总流量,算法具备有效性.

Multi-objective ant colony optimization algorithm for virtual machine placement

The virtual machine placement schemes for existing data centers are mostly concentrated on optimization of energy consumption and resource waste. However, the optimization of datacenter network traffic was rarely considered, which may affect the network scalability. Therefore, with the consideration of both resource waste and total network traffic, this paper models the virtual machine placement as a multi-objective optimization problem, which optimizes the following two objectives in one time for data centers, i.e., minimizing physical machine resources to improve the physical machine efficiency and minimizing total network traffic to improve the network scalability. To solve this problem, we have designed a virtual machine placement algorithm based on multi-objective ant colony optimization (MOACO). Experimental results show that the proposed algorithm can effectively reduce physical machine resources waste and total network traffic compared with the FFD algorithm.