Research Area:  Cloud Computing

Abstract:

Cloud computing facilitates execution of large-scale scientific workflow applications by providing heterogeneous virtualized resources that can be provisioned dynamically. The proliferation of cloud data centers has introduced a serious challenge of enormous energy consumption. Hence, the key concern for executing performance-driven workflows is to devise a scheduling heuristic which can improve utilization of cloud resources and thus helps to reduce power dissipation while adhering to the user’s Quality of Service (QoS) demands. In this paper, we propose a scheduling approach aiming for optimization of makespan, resource utilization, and energy consumption under given deadlines. The proposed heuristic D-DEWS exploits the novel policy of list-based scheduling incorporated with dynamic voltage and frequency scaling (DVFS). It scales the discrete operating frequencies of resources to minimum possible levels, such that a given workflow gets completed within a user-defined deadline with minimum energy consumption. Simulation conducted with synthetic workflows demonstrates the proficiency of proposed heuristic in achieving a significant trade-off between energy savings and performance with deadline compliance. The results obtained confirm that the proposed heuristic outperforms other state-of-the-art algorithms.

Keywords:  

Author(s) Name:  Shalu Saharawat & Mala Kalra

Publisher name:  Springer

DOI:  10.1007/978-981-15-3020-3_33

Volume Information:  volume 3,(2020)

Paper Link:   https://link.springer.com/chapter/10.1007/978-981-15-3020-3_33