Enhanced Iterated Local Search for Scheduling of Scientific Workflows

Abstract

Recently, cloud computing has been used to host and implement scientific applications that have large data and are processing-intensive. This is because cloud computing provides a cost-effective, scalable, and pay-per-use deployment environment. Optimizing workflow scheduling is one of the areas of research activities currently and one of the most important issues in the cloud environment and this problem is considered NP-complete. There are many improvement objectives in workflow scheduling, one of the most important of which is to reduce the makespan. The iterated local search (ILS) has been shown as an effective approach for tackling numerous real-world difficult problems as it simply adapted to different research instances. In this paper, an enhanced ILS algorithm is proposed for achieving this aim. Several enhancing modifications have been proposed for developing the ILS algorithm; two methods for developing an initial solution of the algorithm, and four proposed methods for developing the perturbation phase. Experiments are conducted on well-known scientific workflows of various sizes and types and with WorkflowSim in order to test the six proposed methods. Experimental results show that some of these enhancing modifications outperform the standard ILS as they obtained better results compared with this ILS