Power-Aware Algorithm for Scheduling Virtual Machines in Homogeneous Cloud Datacenters

The ever-increasing utilization of cloud infrastructures has resulted to increased power consumption among datacenters and the accompanying pollution generated by them. In line with green computing initiatives that foster sustainable use of environment-friendly computing facilities, there is a need to address the problem of high rate of pollution generated by datacenters. Several approaches to this problem involve load balancing and scheduling algorithms. The overall objective is to optimize the hosting of Virtual Machines (VMs) in the cloud so as to maximize the utilization of hosts and decrease overall power consumption. In this paper, techniques such as Dynamic Voltage Frequency Scaling (DVFS), fan control, active cooling control and shutdown of underutilized hosts were considered. The proposed algorithm schedules VMs based on their virtual loads among candidate hosts while maintaining the highest overall energy efficiency in small, medium and large homogeneous datacenters. Simulation results demonstrated an improvement in the performance of the previous algorithm under varying parameter values.