Abstract: Concerning the management of the resources in cloud, an energy model and four Virtual Machine Placement (VMP) algorithms were proposed. Firstly, the algorithms calculated the load of the servers in each rack and classified them according to the thresholds. Then the proposed methods selected appropriate virtual machines from servers to migrate using the minimization of migrations policy and accepted the new coming virtual machine allocation requests. At last, the algorithms matched each virtual machine with the servers and found the best server for placement. The experimental results show that, compared with the existing energy-aware resource management algorithm, the proposed technique improves energy efficiency of servers, network equipment, and cooling systems by 2.4%, 18.5%, and 28.1% respectively, resulting in a total of 14.5% improvement on average in the entire datacenter.

Key words: cloud computing, Virtual Machine Placement (VMP), datacenter, threshold, energy efficiency, energy-aware

摘要： 针对云计算的资源管理问题，提出了云计算数据中心的能量模型以及四个虚拟机放置算法。首先计算每个机架上主机的负载并根据设定的阈值进行归类，然后采用最少迁移策略从主机上选择合适迁移的虚拟机并且接受新的虚拟机分配请求，对每个虚拟机与主机集合进行匹配，选择最优化的主机进行放置。实验结果表明，与现有的能量感知资源分配方法相比，该方法在主机、网络设备以及冷却系统方面能量利用率分别提高了2.4%，18.5%和28.1%，总的能量利用率平均提高了14.5%。

关键词: 云计算, 虚拟机放置, 数据中心, 阈值, 能量利用率, 能量感知