Multi-objective ACO virtual machine placement in cloud computing environments

Abstract

Cloud computing systems provide services to users based on a pay-as-you-go model. The more services that data centers deliver to users, the more those centers need to be prepared. However, data centers consume huge amounts of energy from the environment. In order to improve data-center efficiency, resource consolidation using virtualization technology is becoming important for the reduction of the environmental impact caused by the data centers. One of the important keys in resource consolidation is the mapping of virtual machines to suitable physical machines, a procedure called virtual machine placement. The present paper focuses on this problem of virtual machine placement and proposes a multi-objective optimization approach to minimize both power consumption and resource wastage and to minimize energy communication cost between network elements within a data center. An Ant Colony Optimization (ACO) algorithm is proposed to obtain a Pareto set for a multi-objective problem. The proposed algorithms are tested using Cloudsim tools. The performances of these algorithms are compared with three well-known single-objective approaches and a multi-objective Genetic Algorithm (GA). The results demonstrate that the proposed algorithms can seek and find solutions that exhibit balance between different objectives. However, ACO is able to And better solutions than GA in terms of our objectives.