An energy efficient clustering-based scheduling algorithm for parallel tasks on homogeneous DVS-enabled clusters

Huge energy consumption is increasingly become a major obstacle factor to the development of cluster technology, therefore, how to effectively reduce energy consumption in the cluster system is a problem very worthy of study. To address this problem, an algorithm called an Energy Efficient Clustering-based Scheduling Algorithm for Parallel Tasks on Homogeneous DVS-Enabled Clusters (ECSTD) is proposed in this paper. This novel algorithm comprehensively considers the energy consumption of the duplication activities in the scheduling process, and two contributions have been made in this algorithm: (1) delay of the network is taken into account during the scheduling for achieving a more realistic simulation environment; (2) energy consumption is taken into account when a duplication activity is being considered, that is, a proper energy threshold is chosen as a selection criteria of copy tasks, comprehensively considering whether the duplication activity will be proceeded or not. In addition, Dynamic Voltage Scaling (DVS) has been demonstrated as one of the most effective low-power system design techniques, so the DVS technology is used in the algorithm proposed in this paper for obtaining a better energy-saving effect. The final results show that this work is fruitful.