Energy-Efficient Resource Management for High-performance Computing Platforms

Abstract

In the past decade, high-performance computing (HPC) platforms like clusters and computational grids have been widely used to solve challenging and rigorous engineering tasks in industry and scientific applications. Due to extremely high energy cost, reducing energy consumption has become a major concern in designing economical and environmentally friendly HPC infrastructures for many applications. In this dissertation, we first describe a general architecture for building energy-efficient HPC infrastructures, where energy-efficient techniques can be incorporated in each layer of the proposed architecture. Next, we developed an array of energy-efficient scheduling as well as energy-aware load balancing algorithms for high-performance clusters, computational grids, and large-scale storage systems. The primary goal of this dissertation research is to minimize energy consumption while maintaining reasonably high performance by incorporating energy-aware resource management techniques to HPC platforms. We have conducted extensive simulation experiments using both synthetic and real world applications to quantitatively evaluate both energy efficiency and performance of our proposed energy-efficient scheduling and load balancing strategies. Experimental results show that our approaches can reduce energy dissipation in HPC platforms without significantly degrading system performance.