Load Balancing in Grid Computing Using Ant Colony Algorithm and Max-min Technique

Abstract

Stagnation is one of the complicated issues in Grid computing systems, which is caused by random arrival of tasks and heterogeneous resources. Stagnation occurs when a large number of submitted tasks are assigned to a specific resource and make it overflow. To prevent this scenario, a load balancing algorithm based on Ant Colony algorithm and Max-min technique is proposed in this paper. In the proposed algorithm, the resource manager of the system finds the best resource for a submitted task according to a matrix that indicates the characteristics of all resources as pheromone values. By choosing the best resource for the submitted task, a local pheromone update is applied to the selected one to reduce the tendency of being selected by onward new tasks. After this assigned task is executed properly, a global pheromone update is performed to renew the status of all resources for the next submitted tasks. To avoid stagnation, a comparison between a predefined threshold and the pheromone value of each resource is performed to keep the number of assigned tasks below this threshold. Due to harmonizing the resources’ characteristics and tasks, the proposed algorithm is able to reduce the response time of the submitted tasks while it is simple to be implemented.