基于优化 AHP 的虚拟机部署与调度策略

摘要

The dynamic change of task in cloud environment causes load imbalance among the physical servers and affects the uti-lization rate of resources.To solve the problem,a strategy based on AHP and escaping mechanism of physical servers was pro-posed to deploy and schedule virtual machines.The strategy classified the virtual machines according to the characteristics of user’s resource requirements.The competition among virtual machines on the same physical server was avoided by the prelimina-ry selection of physical server through the escape mechanism.The resource of virtual machines was quantified and analyzed based on analytic hierarchy process.The eligible physical servers were forecasted according to the usage of physical server resources. The experimental results show that the strategy can further reduce the number of migration and the occupation of resources in the process of physical server dynamic load balancing,and make the usage of resource among physical servers be more balanced, which improves the utilization rate of resources.%云环境中任务的动态变化会造成物理服务器之间的负载不均衡，影响物理资源利用率。针对这一情况，提出一种层次分析与物理服务器躲避机制相结合的虚拟机部署与调度策略。按照用户服务需求类型对虚拟机进行分类，采用物理服务器躲避机制对目标物理服务器进行初步选择；通过层次分析确定虚拟机内存、CPU 及网络对目标物理服务器选择的影响权重；结合目标物理服务器节点的资源使用情况对该物理节点预测评估。实验结果表明，该策略进一步减少了物理服务器动态平衡过程中的迁移次数以及对资源的占用，使物理服务器之间的资源使用更趋于平衡，提高了物理资源利用率。