Distributed Scheduling on Utility Grids

Grid computing aggregates the power of widely distributed resources and provides non-trivial services to the users. Resource management and scheduling play a crucial role in Grid computing environment. It becomes more challenging in Utility Grids, where consumer wants to pay minimum amount and providers want to earn maximum amount. Many existing scheduling approaches are dedicated to either single criterion or multi-criteria. Existing multi-criteria systems give weight to each criterion according to its relative importance. Users need not worry about the choice of these criteria. These algorithms are less efficient in terms of time complexity and memory utilization. In this paper, we propose an efficient distributed scheduling algorithm on Utility Grids based on Technique for Order Preference by Similarity to Ideal Solution (TOPSIS). We have introduced a new parameter called Processing Element (PE) weight. The objective of this parameter is to select the resource that carves out smallest possible free PE cluster that satisfies the user requirements. User and system defined weights are used to select resource and evaluate the performance of the algorithm. The GridSim toolkit with standard workload model is used to simulate the Grid environment. The simulation results show that our algorithms perform better over existing approaches in terms of user satisfaction and number of application failures.