Simulated Annealing based Heuristic Approach for Dynamic Load Balancing Problem on Heterogeneous Distributed Computing System

Abstract

Load balancing problem on Heterogeneous Distributed Computing System (HDCs) deals with allocation of tasks to computing nodes, so that computing nodes are evenly loaded. Due the complexity of dynamic load balancing problem majority of researchers uses heuristic algorithm to obtain near optimal solutions. We have used consistent ETC (Expected Time to Compute) matrix in to study the performance of simulated annealing (SA) algorithm to minimize the makespan. Simulated annealing based resource allocation algorithm uses sliding widow techniques to select the tasks to be allocated to computing nodes after number of iterations. A new codification suitable for simulated annealing algorithm has been introduced for dynamic load balancing on HDCS. We have also presented three algorithms for move sets representations for SA. Several simulations run have been made on proposed simulated annealing algorithm for dynamic load balancing on HDCS, and compare with conventional first fit (FF), and randomized algorithm. The effect of simulated annealing based dynamic load balancing scheme has been on comparisons with first-fit, and randomized heuristic algorithm with task scalability.