Abstract
Although cloud computing greatly utilises virtualised environments for applications to be executed efficiently in low-cost hosting, it has turned energy wasting and overconsumption issues into major concerns. Cloud infrastructure is built on a great amount of server equipment, including high performance computing (HPC), and the servers are naturally prone to failures.
In this paper, we report on an energy optimisation approach for scheduling HPC applications, applied to decentralised clouds system, that takes dataset transmission energy into account. The optimisation supports combining two conflicting objectives: minimising energy consumption in conjunction with the avoidance of application deadline violations caused by resource failures. Furthermore, we propose two decision strategies for weighing these conflicting objectives dynamically to account for their significance towards producing an ideal energy efficiency and resource utilisation.
Through our developed simulation and experimental analysis using real parallel workloads from large-scale systems, the results illustrate that our approach provides promising energy savings with acceptable level of resource reliability.