中文摘要：

虚拟机迁移是解决物理服务器过载的主要手段之一。作为虚拟机迁移领域的经典算法之一,基于最小迁移时间的虚拟机选择算法虽然具有较低的服务等级协议违例率,但是往往需要迁移多个虚拟机才能解决一次过载,效率较低。为解决这一问题,提出了一种基于容量感知和负载特征的虚拟机选择算法。该方法利用最少虚拟机策略和最小迁移时间策略选择虚拟机,以此降低解决服务器过载时迁移的虚拟机次数和耗费的时间。与基于最小迁移时间的算法相比,文中方法可以在保证服务质量的同时将迁移次数降低56.8%。

英文摘要：

Virtual machine migration is one of the main solutions to deal with overloaded servers. As one of the classic algorithms in the migration, the virtual machine selection algorithm based on minimum migration time has low service level agreement violation. However, it always migrates more than one virtual machines before releasing the overloaded server, which is inefficient. To handle this problem, this paper proposed a virtual machine selection algorithm based on capacity detections and load characters. It adopted minimum migration time and minimum virtual machine policy to reduce the total number of migrated virtual machines and the time consumed. Compared with the algorithm based on minimum migration time, the algorithm can reduce the total number of migrated virtual machines by 56.8% while keeping a high level of adherence to the service level agreements.