Load balancing in grid computing: Taxonomy, trends and opportunities

Grid computing is used to provide different services to users through resources that are geographically dispersed, dynamic, and heterogeneous in nature. In grid computing, load balancing plays a vital role in the reallocation of user jobs when the grid resources become overloaded. In the past few years, scores of load balancing strategies have been proposed by researchers to improve response time, communication overhead, throughput, and resource utilization. In this paper, surveyed load balancing strategies are divided into two broad categories, some supporting task migration and some of them having no support for task migration during the load balancing process. Each category is further categorized based on the basis of grid resource topology, that includes the flat resource topology and hierarchical resource topology. We discussed and compared the number of dynamic load balancing strategies related to these categories on the basis of different load balancing features and performance metrics.