Pareto Based Virtual Machine Selection with Load Balancing in Cloud Data Centre

Abstract

Abstract Cloud Data centers have adopted virtualization techniques for effective and efficient compilation of an application. The requirements of application from the execution perspective are fulfilled by scaling up and down the Virtual Machines (VMs). The appropriate selection of VMs to handle the unpredictable peak workload without load imbalance is a critical challenge for a cloud data center. In this article, we propose Pareto based Greedy-Non dominated Sorting Genetic Algorithm-II (G-NSGA2) for agile selection of a virtual machine. Our strategy generates Pareto optimal solutions for fair distribution of cloud workloads among the set of virtual machines. True Pareto fronts generate approximate optimal trade off solution for multiple conflicting objectives rather than aggregating all objectives to obtain single trade off solution. The objectives of our study are to minimize the response time, operational cost and energy consumption of the virtual machine. The simulation results evaluate that our hybrid NSGA-II outperforms as compared to the standard NSGA-II Multiobjective optimization problem.