基于效益最大化的云虚拟机资源分配研究

Research of virtual machine resources allocation in cloud environment based on utility maximization

针对云资源分配研究中缺乏对用户使用资源的效益情况进行研究的问题,提出了一种使用户效益最大化的云虚拟机资源分配模型。模型建立时以计算资源为研究对象,借鉴了网络带宽分配效用最大化的(network utility maximization,NUM)模型;在分析分配模型时,通过拉格朗日函数将模型简化为求解拉格朗日对偶函数,并引入模糊次梯度算法在理论上证明了可以得到模型的最优解;最后,将模型应用到云虚拟机资源的分配中,并与LINGO软件的结果进行了对比,验证了方案的可行性和算法较好的收敛性。实验结果表明,模型能够很好地实现虚拟机资源分配中用户效益最大化的目标。

In allusion to the actuality in the research of cloud resource allocation that the research based on users5 utilitymerely take up a small part,this paper proposed a cloud allocation model,which maximized the users5 utility. It took the computingresources as the research object and borrowed the idea of NUM model when constructed the model. Then it simplified themodel to Lagrangian dual function by using Lagrange function when analyzed the allocation model and introduced a fuzzy subgradientalgorithm to prove that the model5 s optimal solution could be solved through the algorithm. At last,it applied this modelin the allocation of virtual machine resources, and compared with LINGO software,and verified the feasibility of the projectand convergence of the algorithm. The results of the experiment show that the cloud allocation model is able to realize the targetto maximize the users’ utility very well.