Data centre energy efficiency

Abstract

A data centre is an important component in any organisation as it plays a key role in its growth and success; it has become the most popular cost-effective platform for hosting large scale applications. While more data centres are being implemented and existing facilities are continuously expanding in order to meet the so ever increasing demand, the global network of data centres has become similar to the electricity grid, yet the comparison fails dramatically when it comes to the matter of energy efficiency and cost. Data centres incur frightening costs and in some case spiral out control as regards to power consumption and cooling. An efficient method for saving energy in data centres is to dynamically adjust the data centre compute capacity resources. Nonetheless, this is a challenging solution as it will require thorough understanding of the hosted applications, the resource demand characteristics and the impact on the service level agreement (SLA). In this paper, we investigate the possibility of providing an intelligent control mechanism that manages compute resource capacity dynamically which reduces data centre energy while meeting the performance requirements by means of simulation and extensive analysis using Google's trace workload real data, we will demonstrate how our proposed approach can achieve significant energy savings while meeting the performance requirements.