A simulation model for grid scheduling analysis and optimization

Scheduling is an important research topic in Grid computing. This is due to the aim of Grids to offer high quality services to members of Virtual Organizations based on the efficient use of the available resources. This goal can be achieved through a good scheduling strategy applied to the local lever (clusters) and global level (entire system) of Grids. Since the scheduling problem is NP complete, we can afford only sub-optimal solutions to this problem. In addition, the highly dynamic behavior of Grid components (users, processes, resources) makes more difficult the finding of a good scheduling solution. In this paper we propose a new approach to solving the scheduling problem by simulation. Using a simulator has the merit to shorten the distance between the real Grid system and the model used for its analysis. In order to transform the simulator into a useful tool able to cope with the Grid dynamicity, we propose a solution that couples the simulator with a Grid monitoring/optimization tool, so that scheduling decisions are taken and used in real time for the next short period.