Fast workflow scheduling for grid computing based on a multi-objective Genetic Algorithm

Task scheduling and resource allocation are two of the most important issues in grid computing. In a grid computing system, the workflow management system receives inter-dependent tasks from users and allocates each task to an appropriate resource. The assignment is based on user constraints such as budget and deadline. Thus, the workflow management system has a significant effect on system performance and efficient resource use. In general, optimal task scheduling is an NP-complete problem. Hence, heuristic and meta-heuristic methods are employed to obtain a solution which is close to optimal. In this paper, workflow management based on a multi-objective Genetic Algorithm (GA) is proposed to improve grid computing performance. In grid computing, task runtime is an important parameter. Thus the proposed method considers a workflow as a collection of levels to eliminate the need to check workflow dependencies after a solution is obtained for the next population. As a result, both scheduling time and solution quality are improved. Results are presented which show that the proposed method has better performance compared to similar techniques.