Performance and energy aware scheduling simulator for HPC:evaluating different resource selection methods

摘要

Today, in an energy-aware society, job scheduling is becoming an important task for computer engineersrnand system analysts that may lead to a performance per Watt trade-off of computing infrastructures. Thus,rnnew algorithms, and a simulator of computing environments, may help information and communicationsrntechnology and data center managers to make decisions with a solid experimental basis. There are severalrnsimulators that try to address performance and, somehow, estimate energy consumption, but there are nonernin which the energy model is based on benchmark data that have been countersigned by independent bodiesrnsuch as the Standard Performance Evaluation Corporation. This is the reason why we have implemented arnperformance and energy-aware scheduling (PEAS) simulator for high-performance computing. Furthermore,rnto evaluate the simulator, we propose an implementation of the non-dominated sorting genetic algorithm-IIrn(NSGA-II) algorithm, a fast and elitist multiobjective genetic algorithm, for the resource selection. With thernhelp of the PEAS simulator, we have studied if it is possible to provide an intelligent job allocation policyrnthat may be able to save energy and time without compromising performance. The results of our simulationsrnshow a great improvement in response time and power consumption. In most of the cases, NSGA-II performsrnbetter than other ‘intelligent’ algorithms like multiobjective heterogeneous earliest finish time and clearlyrnoutperforms the first-fit algorithm. We demonstrate the usefulness of the simulator for this type of studiesrnand conclude that the superior behavior of multiobjective algorithms makes them recommended for use inrnmodern scheduling systems.