Load Balancing in Two Types of Computational Environments

Abstract:

In many distributed systems it is possible to share the processing capabilities among the nodes. To accomplish this goal, a number of load balancing algorithms have been proposed in the literature. The purpose of this type of algorithm is to redistribute the system workload with the objective of equalizing the load at each node. In this report we discuss two types of distributed computing environments for which load balancing techniques yield increased performance: (1) pool of processors and (2) independently owned processors. Both environments are built around a loosely-coupled computer network, but in the former all computational nodes belong to one user community, while in the latter each machine belongs to a different individual or organization. We argue that these two environments should not be treated with the same type of load balancing algorithm due to their distinct nature. For the first environment
we discuss a load balancing algorithm based on executing jobs at the least loaded processor in the network. For the second environment we introduce the concepts of lending and borrowing resources. For both environments we present empirical results to illustrate our approach to load balancing.