多租户云计算中基于ILP模型的虚拟机放置策略  

Virtual Machine Placement Strategy Based on ILP Model in Multitenant Cloud Computing Environment

作　　者：王准 苏顺开[2] 

机构地区：[1]广州工商学院计算机科学与工程系,广东广州510850 [2]中山大学现代教育技术研究所,广东广州510275

出　　处：《湘潭大学自然科学学报》2016年第4期71-75,共5页Natural Science Journal of Xiangtan University

基　　金：广东省青年创新人才类项目(2015KQNCX196);2016年广东省高等教育学会高职高专云计算与大数据专业委员会课题(GDYJSKT16-06)

摘　　要：针对云计算中多个租户租用计算资源时,分布式数据中心中虚拟机(VM)的优化放置问题,提出一种基于整数线性规划(ILP)模型的VM放置策略.首先,以最小化数据中心之间的通信量为目标,在考虑VM位置和数据中心容量等约束下,构建一个ILP优化模型.然后,提出一种变量聚合方法来简化ILP模型,减少变量数目,提高计算效率.最后,通过求解简化ILP模型来获得VM的最优放置方案.实验结果表明,该方案能有效降低数据中心间的数据通信量,从而降低了通信成本,同时具有较低的计算时间.For the issue that the placement optimization of virtual machine（VM）in distributed data centers when multiple tenants lease computing resources,a VM placement strategy based on integer linear programming（ILP）model is proposed.Firstly,to minimize the traffic between data center as the goal,a ILP optimization model is constructed by considering the constraints of VM location and data center capacity.Then,a variable aggregation method is proposed to simplify the ILP model,to reduce the number of variables,and improve the computational efficiency.Finally,the optimal placement scheme of VM is obtained by solving the simplified ILP model.Experimental results show that the proposed scheme can effectively reduce the amount of data communication between DC,which can reduce the communication cost,and has lower computation time.

关 键 词：云计算 多租户 虚拟机放置 整数线性规划模型 最小化通信量 

分 类 号：TP312[自动化与计算机技术—计算机软件与理论]