Performance Analysis of Cloud Systems with Load Dependent Virtual Machine Activation and Sleep Modes

Abstract

Due to the advancements in virtualization technology, it is now an up and coming field and has become a more appealing area of internet technology. Since there is a rapid growth for the demand of computational power increases by scientific, business, and web-applications, it leads to the creation of large-scale data centers. These data centers consume enormous amounts of electrical power. In this article, the authors study energy saving methods by consolidation and by switching off those virtual machines which are not in use. According to this policy, c virtual machines continue serving the customer until the number of idle server attains the threshold level d; then d idle servers take synchronous vacation simultaneously, otherwise these servers would begin serving the customers. Numerical results are provided to demonstrate the applicability of the proposed model for data center management in particular, to quantify the tradeoff theoretically between the conflicting aims of energy efficiency and QoS.