Scheduling methods of virtual machine (VM) migration are regarded as an effective way for energy conservation. A number of scheduling methods of VM migration have been proposed and/or improved in several research works. However, none of the scheduling methods was evaluated from a comprehensive viewpoint, and there is no useful reference for practices of energy conservation in various data centers. In order to provide a useful reference for best practices of energy conservation in various data centers, this paper presents the first comprehensive evaluation of scheduling methods of VM migration. After giving an overview of major optimization problems, this paper proposes a new set of evaluation metrics that can be used in the evaluation of various scheduling methods of VM migration from different aspects, presents an evaluation environment we constructed according to our evaluation metrics, and discusses evaluation results of those proposed scheduling methods on our evaluation environment. The results of our comprehensive evaluation show that the proposed evaluation metrics can reflect the merits and demerits of various scheduling methods from the perspective of different data center scales and divergent workload types and therefore can be used for evaluating the scheduling methods of VM migration fairly and reasonably.

No abstract