A comparative study on performance of energy efficient load balancing techniques in cloud

Abstract

The advent of cloud computing has radically changed the development of the future Internet of Services. Cloud data centers accommodating numerous tenant requests for cloud applications discharge massive quantities of energy, contributing to high operational expenditures and carbon dioxide (CO <inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</inf> ) diffusion into the environment. In order to curtail this, there is a need to conserve energy for future use, by implementation of a new energy efficient mechanism and observe its results in a cloud data center. The pruned energy consumption will subsequently abate the cost of computing resources. An effective means to save energy conservation to achieve energy efficient data center in cloud is to adapt to optimal load balancing techniques, which tries to improve the performance by evenly distributing the workload to minimize the enormous energy consumption of overloaded servers in cloud data centers. The crux of this work is to trigger research and development of models by comparing and evaluating the performance of various load balancing techniques in cloud to find out which technique is more energy efficient so as to aid in mitigating energy consumption by frequently dispensing the load and improves the overall performance of the system under consideration. A comparative analysis in terms of performance is carried out for various load balancers like Weighted First-cum-First-Served, First-cum-First-Served, Round Robin and Throttled load balancer, by considering metrics like overall Energy consumption, Processing and Execution time and Operating and Processing cost. Thus, this enables to achieve an overall energy efficient cloud data center and the experimental results show that Weighted First-cum-First-Served (WFCFS), which is a self-adaptive and dynamic algorithm, improves the overall performance by devouring less time for scheduling virtual machine and allocates the incoming random requests to all the available virtual machines in an efficient manner and thereby achieves minimum power consumption, minimum cost, minimum overall response time and optimal throughput, in comparison with other load balancers.