Power-aware scheduling of real-time virtual machines in cloud data centers considering fixed processing intervals

Abstract

In Cloud data centers, Virtual Machines (VMs) as resources for Infrastructure as a service (IaaS) can be dynamically allocated to different customers to meet their application goals. In this paper, from the providers' point of view of reducing power consumption, we investigate the power-aware scheduling of real-time VMs by considering fixed processing intervals. In the case of all VMs sharing random portions of the total capacity of a Physical Machine (PM), finding the optimal solution of minimizing the total number of PMs is NP complete as proved in many open literature. Hence, we model the problem as a modified interval partitioning problem to provide approximate solutions and propose scheduling schemes to reduce the power consumption. Simulation results show our proposed scheduling schemes incur 8%-40% less power consumptions than existing algorithms.