提高云计算系统资源利用率一直是云计算研究的重点内容之一。基于此,将传统的多目标蚁群算法进行改进,并结合排除法解决虚拟机放置的多目标优化的问题,该算法可以通过信息素的不断更新,最终收敛得到最优解。主要考虑了服务级合约违背率(S)、资源损耗(W)、电源消耗(P)3个因素。实验结果表明,与传统的启发式方法和遗传算法相比,该算法有利于并行计算,能够在多个相互冲突的目标间实现最优权衡和折衷,在服务级合约违背率较少的情况下,系统资源浪费和电源消耗最少,具有可行性。
Improving the resource utilization of cloud computing system has been one of the important content of cloud computing research. Based on this purpose, the traditional multi-objective ant colony algorithm was improved, and com- bined with the exclusive method to solve the problem of multi-objective optimization of virtual machine placement, the al- gorithm can through the pheromone updating to finally get the optimal solution convergence. In this paper, we considered three factors: the service level agreement violation rate (S), and the resource depletion (W), the power consumption (P). The experimental results show that compared with traditional heuristic method and genetic algorithm, the algorithm is conducive to parallel computing and can be implemented in a number of conflicting goals between the optimal balance and compromise, In the case of service level agreement violation rate less, waste of the resources of the system and power con- sumption are both minimum.