Optimized scheduling of scientific workflows based on iterated local search

Abstract

Recent years have witnessed a great interest in scientific applications with large data and processing-intensive, so cloud computing is used which provides the resources needed to implement and run these applications. One of the challenges in the management of scientific workflow applications is scheduling them to solve many combinatorial optimization problems, including reducing execution time, cost, resource utilization, and energy consumption. Due to the fact that the iterated local search algorithm (ILS) has been successfully applied to solve many combinatorial optimization problems, this paper investigates the performance of ILS in solving the scientific workflow scheduling problem which is a highly constrained problem. The main components that are different from one problem to others are the ILS parameters, local search, and perturbation, which must be carefully designed. The performance of the standard ILS has been examined and compared with the latest technology. The experimental results show that the proposed algorithm (ILS) obtained good results compared to the best-known results in the literature. This is due to the ILS being an adaptable metaheuristic, which can be simply adapted to different search situations and instances.