Energy-performance trade-off through restricted virtual machine consolidation in cloud data center

Abstract

Cloud, the current boom in information and communication technology, has forced the new organizations to think twice before setting up their own infrastructure as it can be rented totally on pay-as-you-go basis from the Cloud Service Providers (CSP). However the large amount of electrical energy consumed by cloud data centers creates one of the major hindrances in the advancement of cloud by increasing the operational costs. It also affects the environment by the emission of carbon dioxide. These challenges motivate the researchers to develop algorithms that can minimize consumption of energy in cloud data center. VM consolidation is one of such methods that saves energy by increasing total count of inactive servers. But its effect on the overall cloud performance is a bit negative. The average response time of the jobs increases since the workload is distributed on lesser number of active servers. So to maintain the energy-performance trade-off VM consolidation must be done in a restricted manner. The paper has proposed a novel approach of adding a constraint to the existing VM consolidation technique to avoid unnecessary VM migration. It saves energy by eliminating the repeated migration of the same VM. We have also proposed heuristics for virtual machine selection algorithm. CloudSim tool is used to simulate the proposed algorithms. Results show that they significantly decrease the consumption of energy and also fulfill the service-level agreement (SLA).