A novel cost-efficient approach for deadline-constrained workflow scheduling by dynamic provisioning of resources

Abstract

Workflow scheduling is a crucial aspect of cloud computing that should be performed in an efficient manner for optimal utilization of resources. The development of a cost-efficient algorithm has always been an important topic of research in this regard. In this paper, we propose a novel workflow scheduling algorithm, which is cost-efficient and deadline-constrained. The proposed algorithm is consolidated by dynamic provisioning of the resources, using k-means clustering technique and a variant of the Subset-Sum problem. In the algorithm, we consider level based scheduling using the concept of Bag of Tasks (bots) and develop a new technique for associating deadlines with each bot. Through extensive simulation runs, we show that the proposed algorithm outperforms the existing algorithms like Dynamic Provisioning Dynamic Scheduling (DPDS) and Infrastructure as a Service (IaaS) Cloud-Partial Critical Path (IC-PCP). The effectiveness of our algorithm over these two algorithms is also illustrated through the popular statistical test ANOVA and its subsequent post-hoc analysis.