An algorithm to select an energy-efficient sever for an application process in a cluster of servers

Abstract

We have to decrease electric energy consumption of information systems, especially servers to reduce carbon dioxide emission. In information systems, a client issues application processes to servers in clusters. Application processes have to be performed on servers so that the total energy consumption of servers in the cluster can be reduced. In this paper, we discuss how to select a server to energy-efficiently perform an application process issued by a client. In order to find an energy-efficient server, we have to estimate the execution time of application processes and the energy consumption of the server to perform but new application processes issued and current application processes. In this paper, we newly propose an algorithm to estimate the execution time of application processes and the energy consumption of a server by considering not only current active application processes but also possible application processes to be issued after the current time. By using the estimation model, we also propose an MES (Minimum-Energy Server selection) algorithm to select a server to perform an application process. We design and implement an EDS (Eco Distributed System) simulator to evaluate selection algorithms in terms of energy consumption of servers and execution time of each process.