Improving Electricity Usage based on Computational Modeling in Cloud Computing

Abstract

The achievement of energy efficiency is gradually receiving a lot of attention these days due to the budget and environmental issues. A prediction technique has been developed in our previous research to improve monitoring statistics. In this research, our new proposal can make the optimization to solve the energy issue of cloud computing by adopting the predictive monitoring information. Actually, the convex optimization technique is coupled with the proposed prediction method to produce a near-optimal set of physical machines hosting. After that, an appropriate migration instruction may eventually be created. The cloud orchestrator can relocate virtual machines to a designed sub-set of infrastructure on the basis of this instruction. The idle physical servers can then be switched off appropriately to save power and maintain system performance. For evaluation purposes, an experiment is conducted based on Google Traces 29-day period. By using this assessment, the proposed approach demonstrates the potential to significantly reduce power consumption without affecting service quality.