An adaptive scheduling framework for the dynamic virtual machines placement to reduce energy consumption in cloud data centers

Abstract

Cloud computing has revolutionized the IT industry through its on-demand provisioning of virtualized resources through the internet. Although it relies on sharing of resources to improve the performance of datacenters, it has increased the complexity of IT systems in recent years. To meet the market requirements, cloud providers are expanding their datacenters with a large number of servers leading to high energy consumption and therefore, increasing the carbon footprint. Environmental impact and rapidly surging energy costs have become a major concern for both the government bodies and the IT service providers. In this paper, we propose a genetic algorithm based hybrid load management strategy which uses multiple existing VM allocation policies to minimize the energy consumption, Service Level Agreement (SLA) violations and number of VM migrations. The presented solution approach is evaluated on CloudSim Plus simulation framework using the well known PlanetLab workload. The results obtained from the experiments show substantial improvement in energy consumption in comparison to the individual approaches while maintaining the performance constraints.