A Multiobjective Resources Scheduling Approach Based on Genetic Algorithms in Grid Environment

Abstract

Resources scheduling plays an important role in grid. This paper converts resources scheduling problem in grid into a multiobjective optimization problem, and presents a resources scheduling approach based on multiobjective genetic algorithms. This approach deals with dependent relationships of jobs, and regards multi-dimensional QoS metrics, completion time and execution cost of jobs, as multiobjective. Based on Pareto sorting and niched sharing method, our approach determines optimal solutions. Experimental results show that our approach gets less completion time of jobs and total execution cost of jobs than min-min algorithm and max-min algorithm