Scientific Workflow Scheduling by Adaptive Approaches with Convex Optimization in Cloud Environment

Abstract

Scientific workflows includes large scale data transfer which consumes resources of a system. Cloud is a dynamic environment where VMs are rented as per need. The scheduling of tasks to a VM in cloud system depends on multi objectives at a time, not on single objective. For scheduling the workflow tasks on VMs, efficient techniques are needed. To make a cloud system an efficient one, its resources need as maximum utilization as possible. If we talk about VMs, Virtual system has made scheduling fast in all aspects yet scheduling needs an efficient way so that total various quality parameters like total execution time, total execution cost, energy consumption as well as response time can be minimized. The aim of this study is to perform tasks migration from less utilized cloud resources to other ones but without increase in total execution time. We proposed a better optimization approach and applied the same on already scheduled tasks to cloud resources. The proposed approach called TBW optimization approach is based on implementation of three optimization techniques named Tabu optimization, Bayesian Optimization and Whale Optimization. In this research, our goal is to improve the utilization of cloud resources as well as enhancing the performance of the cloud system. The experiments using different scientific workflows highlights that TBW optimization highlights the effectiveness and better results than existing approaches like GA-PSO, Whale optimization, GA and PSO.