Optimized Grid Scheduling Through Adaptive Mutation-Based Swarm Intelligence

Abstract

The grid computing model has proved to be a successful paradigm for running large-scale distributed applications, using geographically distributed computing resources. But scheduling of tasks efficiently is still a major challenge as the resources are heterogeneously organized, the workloads are dynamically changing and the requirements for executing a task have also changed. The traditional scheduling methods have limitations of being flexible and making effective use of resources in a complex grid system. The present paper introduces an Optimized Grid Scheduling framework which is based on Adaptive Mutation Based Swarm Intelligence (AMSI) for gaining scheduling efficiency and resource allocation. The proposed framework combines swarm intelligence with an adaptive mutation mechanism to adapt the exploration and exploitation capabilities at runtime of a scheduling process. Structured representation is used to model grid resources and tasks; intelligent population initialization improves the convergence towards high quality scheduling solutions. Various workload scenarios are used to run the extensive simulations to test the proposed approach. The performances are evaluated based on makespan, resource utilization, throughput and load-balancing efficiency measures. Experimental results show that the AMSI framework always outperforms traditional scheduling algorithms such as FCFS, Round Robin, Opportunistic Load Balancing and the standard Particle Swarm Optimization. The proposed method can be applied to the grid computing environment and has the advantages of better scheduling quality, faster convergence and better resource sharing.