Incorporating Migration Control in VM Selection Strategies to Enhance Performance

In this work we have designed three algorithms, by doing amalgamation of heuristic method and migration control. Virtual Machine (VM) selection is the immediate step after the host overload is detected. Now decision needs to be made to choice the VM(s) to migrate. VM selection is an important task and the energy consumption depends on the selection. Three VM selection strategies have been proposed in literature and been incorporated in CloudSim, a simulation framework to simulate Cloud architecture, resource provisioning etc. Migration control could influence the VM selection mechanism as steady and resource consuming VMs not to be selected. Reason behind is that, a steady VM which is continuously consuming recourses has high probability of consuming resource in the same manner in future. So if we migrate such VM it will also overload the next datacenter in which we are migrating it to. In this proposed model, CPU usage has been considered to identify a VM to be steady or not. The VM selection mechanisms are modified and equations are also modified to adopt migration control in the policy and algorithms also developed. We have used Cloudsim toolkit to simulate our experiment and to evaluate the performance of the proposed algorithm, we have used real-world work load traces of Planet lab VMs. From the comparison with existing energy aware VM consolidation methods, it is found that performance of our algorithm is better than other proposed heuristic approaches.