A learning‐based approach for virtual machine placement in cloud data centers

Summary In recent years, the increasing use of cloud services has led to the growth and importance of developing cloud data centers. One of the challenging issues in the cloud environments is high energy consumption in data centers, which has been ignored in the corporate competition for developing cloud data centers. The most important problems of using large cloud data centers are high energy costs and greenhouse gas emission. So, researchers are now struggling to find an effective approach to decreasing energy consumption in cloud data centers. One of the preferred techniques for reducing energy consumption is the virtual machines (VMs) placement. In this paper, we present a VM allocation algorithm to reduce energy consumption and Service Level Agreement Violation (SLAV). The proposed algorithm is based on best‐fit decreasing algorithm, which uses learning automata theory, correlation coefficient, and ensemble prediction algorithm to make better decisions in VM allocation. The experimental results indicated improvement regarding energy consumption and SLAV, compared with well‐familiar baseline VM allocation algorithms.