Computational Intelligence Techniques for Energy Management in Cloud Data Centers

Abstract

AbstractEnergy management in cloud datacenters attracts significant attention due to the high power consumption of datacenter. The cloud datacenter consumes a large amount of energy while transmitting and computing the data applications, which increases the maintenance cost for cloud datacenters. In order to solve these issues, several optimization techniques have been applied for load balancing. This paper details various computational intelligence methods used for energy management in cloud datacenters with its advantages and limitations. The three major aspects for energy management in cloud datacenters are load balancing, resource allocation, and predictive models. Machine learning techniques were used in the prediction of load and applying optimization technique to prevent service-level agreement (SLA) violation. Hybrid fruit fly optimization method, genetic algorithm, and multi-objective grey wolf optimization methods shows higher efficiency in minimizing energy consumption in datacenters. The hybrid method of firefly and improve particle swarm optimization (PSO) provides the solution with less computational time. Many existing methods suffer from the problem of local optimum and poor convergence along with SLA violations while considering energy efficiency.KeywordsComputational intelligenceCloud datacentersEnergy managementLoad balancing and predictive models