Energy-Aware Processor Merging Algorithms for Deadline Constrained Parallel Applications in Heterogeneous Cloud Computing

Abstract

Energy efficiency has become a key issue for cloud computing platforms and data centers. Minimizing the total energy consumption of an application is one of the most important concerns of cloud providers, and satisfying the deadline constraint of an application is one of the most important quality of service requirements. Previous methods tried to turn off as many processors as possible by integrating tasks on fewer processors to minimize the energy consumption of a deadline constrained parallel application in a heterogeneous cloud computing system. However, our analysis revealed that turning off as many processors as possible does not necessarily lead to the minimization of total energy consumption. In this study, we propose an energy-aware processor merging (EPM) algorithm to select the most effective processor to turn off from the energy saving perspective, and a quick EPM (QEPM) algorithm to reduce the computation complexity of EPM. Experimental results on real and randomly generated parallel applications validate that the proposed EPM and QEPM algorithms can reduce more energy than existing methods at different scales, parallelism, and heterogeneity degrees.