Using Artificial Neural Network for VM Consolidation Approach to Enhance Energy Efficiency in Green Cloud

Abstract

Cloud computing is a popular on-demand computing model that provides utility-based IT services to the users worldwide. However, the data centers which host cloud applications consume an enormous amount of energy contributing to high costs and carbon footprints to the environment. Thus, green cloud computing has emerged as an effective solution to improve the performance of cloud by making the IT services energy and cost efficient. Dynamic VM consolidation is one of the main techniques in green computing model to reduce energy consumption in data centers by utilizing live migration and dynamic consolidation. It minimizes the consumption of energy by monitoring the utilization of resources and by shifting the idle servers to low power mode. This paper presents a VM selection approach based on artificial neural network (ANN). It uses backpropagation learning algorithm to train the feedforward neural network to select a VM from an overloaded host. Thus, it optimizes the problem of VM selection by learning training dataset and enhances the performance of selection strategy. To simulate our proposed algorithm, we have used MATLAB and the simulation result depicts that our proposed method minimizes the energy consumption by 30%, SLA violation by 3.51%, the number of migrations by 10%, and execution time by 29.7%.