A Table Driven Dynamic Load Balancing Scheme for Distributed System

In a distributed network of different computing hosts, the performance of the distributed system depends on how the work is efficiently divided among all the participating nodes. Load balancing is an important activity wherein tasks are assigned to participating nodes with a view to minimize the waiting time and bring improvement in execution time of the system. The existing load balancing techniques consider each participating node having equal computing capability .In this work we propose a load distribution algorithm for a cluster of heterogeneous commodity hardware. The algorithm considers the nature of applications at hand before allocating the computing resources. A considerable improvement in execution time of parallel applications has been observed especially when tasks are sent and processed on favorable nodes instead of randomly assigned nodes.