The Reasearch On Virtual Machine Migration Oriented To Energy Consumption In Cloud Computing

Cloud computing, as a new model of service, is usually composed by millions of different physical devices. Using virtualization technology, Cloud computing can provide users with virtualizaed resources that can be requested at any time and be paid on-demand. However, the energy consumption of the data center is grown exponentially followed by the expanding of the data center’s scale. Therefore, how to reduce the energy consumption of the whole data center by the virtual machine migration has become the current focus of many studies.The virtual machine migration has three main processes: host overloaded detecting, virtual machine selection and virtual machine placement. Researches on the three processes have been conducted to solve the problem. At first, during the virtual machine selection process, this paper proposes an algorithm called a VM selection algorithm based on the degree of the performance satisfaction, which calculates the level of performance satisfaction of each virtual machine on the overloaded hosts. After that, a policy that picks out the target host on which the migratable virtual machine has the least influence after the placement is designed. That policy is called the virtual machine placement algorithm based on the minimum correlation coefficient and short for MC. However, MC only takes the CPU utilization into consideration. In fact, there are some other resources that has the great impact on the performance of the whole data center, such as memory, network, disk and so on. Therefore, an improved VM placement algorithm called predicted affinity based virtual machine placement in cloud computing environment is proposed. The character of the algorithm is that it has no limits of the resources’ dimensions and it introduces a new concept: affinity, which is used to quantify the relationship between the virtual machines and hosts.Finally, we have designed a virtual machine migration system based on the OpenStack platform and the VM selection and placement algorithm have been implemented in this system. The experiment results show that the algorithms have a certain effect on the reduction of the energy consumption, VM migration times and the SLAs.