A new reliability‐based task scheduling algorithm in cloud computing

Abstract

SummaryIn the last decade, the scale of heterogeneous computing (HC) systems such as heterogeneous cloud computing environments was growing like never before. So network failures are unavoidable in such systems, which affect system reliability. Since the task scheduling algorithm in HC is challenging, we investigate a new reliability‐aware task scheduling algorithm (RATSA) in this paper. RATSA is designed to schedule tasks on directed acyclic graphs (DAGs) by using the shuffled frog‐leaping algorithm (SFLA) and genetic algorithm (GA) as evolutionary algorithms. The population‐based SFLA‐GA is applied to optimize makespan in the RATSA as an NP‐complete problem. Moreover, the proposed algorithm exploits a new heuristic‐based earliest finish time technique for task mapping to virtual machines (VMs) section to decrease the failure rate. Experimental results on random DAGs indicate that RATSA improves some current algorithms in terms of reliability and has acceptable performance in makespan. The results reveal that the RATSA decreases the overall failure rate by 43% compared to some current task scheduling algorithms.