A Network and Power Aware framework for data centers using virtual machines re-allocation

Abstract

Many recent studies have proposed energy efficient approaches that use various techniques such as Dynamic Voltage Scaling (DVFS), CPU idling, and server consolidation in data centers. Also, many networking problems could be introduced by server consolidation techniques. In this research, we introduce an enhanced technique called Network and Power Aware (NPA) server consolidation that will dynamically reallocate VMs between hosts in a manner that reduces the power consumption, conserves bandwidth, mitigates the traffic congestion at the uplinks of the switches and routers, and even enhances the overall network performance of cloud data centers. We used Cloudsim as a simulator in our evaluation. We compared our proposed algorithm with DVFS and Static Threshold VM reallocation with Random selection (ThrRs) techniques based on SLA violations due to network performance, the total throughput of the data center, total number of migrations and the total power consumed. The results confirm that our proposed approach conserves power, resolves networking problems and enhances the overall network performance.