Energy-Aware Resource Selection for Asynchronous Replicated System in Utility-based Computing Environments

This paper addresses the problem of energy-aware resource selection for applications that update data in utility-based computing systems. The problem is complex to be solved, especially in asynchronous replicated system, since one should first provide a good prediction scheme on the performance evaluation of resource selection model in the presence of job that update data before the energy-efficiency issue can be addressed. With this in mind and by observing that simplicity implies efficiency and scalability, two approaches of energy-aware resource selection are proposed and developed. The main objective is to minimize the energy consumption while still preserves a minimum system performance degradation trade-offs. The proposed approaches exploit both the presence and the absence of knowledge on resource energy usage to select the best resources to run the job. Also, they take advantage on the nature of high performance of job execution in asynchronous replicated system in its resource selection decision making process. The experimental results proved that the proposed approaches are able to minimize resource energy usage while preserving minimum performance degradation trade-offs.