An effective multi-objective workflow scheduling in cloud computing: A PSO based approach

Abstract

Cloud computing has emerged as prominent paradigm in distributed computing which provides on-demand services to users. It involves challenging areas like workflow scheduling to decide the sequence in which the applications are to be scheduled on several computing resources. Due to NP-complete nature of workflow scheduling, finding an optimal solution is very challenging task. Thus, a meta-heuristic approach such as Particle Swarm Optimization (PSO) can be a promising technique to obtain a near-optimal solution of this problem. Several workflow scheduling algorithms have been developed in recent years but quite a few of them focuses on two or more parameters of scheduling at a time like usage cost, makespan, utilization of resource, load balancing etc. In this paper, we present a PSO based workflow scheduling which consider two such conflicting parameters i.e., makespan and resource utilization. With meticulous experiments on standard workflows we find that our proposed approach outperforms genetic algorithm based workflow scheduling in all cases achieving 100% results.