Performance of an IaaS cloud with live migration of virtual machines

Cloud centers often use virtualization which requires live migration of virtual machines to improve performance and availability. In this paper, we describe an analytical performance model that measures the important performance indicators, namely, rejection probability and total delay, of a cloud center whilst taking into account live migration of virtual machines. Using the proposed performance model, we show that live virtual machine migration reduces task rejection probability and renders the super-task delay nearly independent of mean service time of individual tasks in a super-task.