Energy-aware resource selection on opportunistic grids

Energy consumption has been a constant concern for high-performance computing (HPC). Recently, this concern has gained attention from the research community, which is aiming to reduce its costs. The performance gain in such an environment is usually proportional to cost. Examples of such environments are computational grids, which are used in the academic and enterprise domains. On the other hand, one way of obtaining high-performance computing with low-cost investment is by using opportunistic grids, which have become a viable alternative to super-computers and dedicated clusters. This paper proposes an energy-aware resource-selection algorithm to reduce energy consumption in opportunistic grids. The proposed algorithm takes into consideration resource status as well as actions to be taken before allocation to calculate energy consumption. Experimental analysis conducted in this study, taking into account network traffic and node status, shows that a more efficient resource-selection outcome can be obtained, leading to reduced energy consumption. Tests demonstrate an energy-consumption reduction of around 9.5% compared to a commonly used approach.