Low Energy Consumption Strategy Of Deployment And Migration For Virtual Machine In Cloud Data-center

In recent years, with the data center’s size rising, the proportion of the electricity consumption incurred against the total cost. High energy consumption has become a major factor limits the development of the data center.To face the energy challenges of data center, the thesis study on the deployment and migration of virtual machines based on the visualization and other energy-saving technologies.The thesis presents low energy consumption virtual machine deployment strategy based on data center model and power function, which includes a virtual machine initial deployment algorithm named BT-MPA and dynamic virtual machine migration algorithm SE-MMA. BT-MPA algorithm can obtain optimal virtual machine deployment using backtracking according to the deployment target. SE-MMA includes non-overload migration and overload migration. In order to aggregate resources, SE-MMA migrate virtual machine with SMP under non-overload state. And when host is under overload, data center use the minimum migrate time policy choice migrated virtual machines, and redeploy these virtual machines by BT-MPA, the minimum time problem is solved by greedy and dynamic programming algorithm.The thesis uses CloudStack to build laaS cloud platform, and designs three sets of simulation. The simulation one compares different deployment algorithms, which show that IT facilities’power consumption has reduced9.7%and the open host number has reduced17.3%by using the BT-MPA algorithm with respect to MBFD, indicating that the BT-MPA algorithm can reduce the number of open physical machines, reducing data center energy consumption;. The experiment two compared different virtual machine selection policy, in which greedy algorithm and dynamic programming algorithm has reduced virtual machine migration time by50%, the result proves greedy and dynamic programming algorithm can solve the minimum migration time problem. Simulation three compared the different migration algorithms, in which SE-MMA algorithm with respect to the MM algorithm reduces the energy consumption by14%, verifying the SE-MMA algorithm can achieve lower energy consumption and achieve the purpose of energy conservation.