In order to reduce the energy consumption of large- scale data centers,a virtual machine (VM) deploy-ment algorithm called TTEA (Three Threshold Energy- saving Algorithm) is proposed.TTEA makes use of the linearrelationship between energy consumption and processor resource utilization,and migrates the VMs on heavily- loadedand lightly- loaded hosts to the host with proper load,while the VMs on normally- loaded host remain constant.Then,on the basis of TTEA,four VM selection strategies are proposed and HLGP is chosen as the best one as aresult of experimental comparison.Simulated results indicate that,in comparison with the single threshold algorithmand the double threshold algorithm,HLGP strategy is more effective because it efficiently reduces the energy con-sumption of data centers and keeps high quality of service.