Energy efficient task scheduling for parallel workflows in cloud environment

Abstract

The demand for the Cloud services are increasing day by day and so the resources in the Cloud data centers. To meet the demands, a lot of research has done in reducing the service response time by increasing the utilization of the resources, but neglected the energy consumption of the resources. The data centers consume huge amount of energy and dissipate carbon footprints in the environment. The energy consumption in Cloud includes the energy consumed by the servers, memory, network, cooling systems and conversion. As the servers consumes major fraction of energy, we consider our work on server's energy consumption. The parallel applications are gaining importance in Cloud that throws a significant challenge in energy saving of the Cloud servers. In this paper we propose a method to reduce the energy consumption by using the Dynamic Voltage Frequency Scaling technique where the servers operate at different levels of voltage by reducing the operating frequency. We use the slack time between the tasks to sacrifice the operating frequency so that the schedule do not violate the deadline of parallel applications. We used the real world applications represented by Directed Acyclic Graphs for simulation purpose. The results shows that the proposed algorithm achieves the significant energy saving with the existing approaches.