Fine Tuning Performance and Efficiency of Virtual Infrastructure

Abstract

This is a study of virtual infrastructure performance. We intend to develop methods to correctly size virtual machine resources according to its workload. We will attempt to optimize virtual machines over four different resource categories: CPU, storage, memory, and network. We will host virtual machines and modify parameters per machine while monitoring its workloads. The results will show performance differences between optimized and default resource allocations. The implications of these results can result in increased cost and energy savings for infrastructure providers.