Greening the Cloud Through Power-Aware Virtual Machine Allocation

Abstract

Cloud computing has tremendous ability to deliver services on time, conditional to host redundancy to guard against failure and excess server capacity while handling spiky demand. This results in a requirement of more servers than actually required and their average utilization is about 10% or less. This leads to huge amounts of power consumption as servers draw nearly the same amount of power regardless of their current utilization. This huge amount of power consumption can be controlled by power-efficient allocation of resources. In this paper, we proposed algorithms for the power-aware allocation and migration of virtual machines. Power saving is achieved through power efficient consolidation of virtual machines on a smaller number of servers and by putting idle nodes in sleeping mode. The decision of virtual machines allocation and consolidation on a server is based on server efficiency, i.e. minimum energy consumption with the maximum utilization. The simulation results show that the proposed method performs better than the recent power efficient approach.