Optimization of the overload detection algorithm for virtual machine consolidation

Abstract

With the increasing demand of computing resources, a lot of large-scale data centers have been established around the world. Data centers provide efficient and convenient computing services while consume enormous energy at the same time. Therefore, reducing the energy consumption in data center is important and meaningful. Virtual machine consolidation (VMC) is an effective strategy that can improve the energy efficiency of data centers. In this research, we have put forward a new overload detection algorithm on the basis of existing algorithms to achieve better tradeoff between energy and service level agreement (SLA) violation. The prediction process of the proposed algorithm is mainly based on the utilization history of VMs. The experimental results demonstrate that compared with the original host-based framework, the proposed VM-based framework reduces the average ESV by 23.4%.