Implementation of Load Balancing Policies in Distributed Systems

OF THESIS Submitted in Partial Fulfillment of the Requirements for the Degree of Master of Science Electrical Engineering The University of New Mexico Albuquerque, New Mexico June, 2004 Implementation of Load Balancing Policies in Distributed Systems by Jean Ghanem B.E., American University of Beirut, 2002 M.S., Electrical Engineering, University of New Mexico, 2004 Abstract Load balancing is the allocation of the workload among a set of co-operating computational elements (CEs). In large-scale distributed computing systems, in which the CEs are physically or virtually distant from each other, there are communicationrelated delays that can significantly alter the expected performance of load-balancing policies that do not account for such delays. This is a particularly prominent problem in systems for which the individual units are connected by means of a shared communication medium such as the Internet, ad-hoc networks, wireless LANs. Moreover, the system performance may greatly vary since it incorporates heterogenous nodes that are not necessarily dedicated to the application at hand. In such cases, an actual implementation becomes necessary to understand the load-balancing strategies and their reactions when employed in several environments since mathematical models may not always capture the unpredictable behavior of such systems. In this thesis we propose a software implementation architecture where several distributed load-balancing strategies could be tested and verified under different en-Load balancing is the allocation of the workload among a set of co-operating computational elements (CEs). In large-scale distributed computing systems, in which the CEs are physically or virtually distant from each other, there are communicationrelated delays that can significantly alter the expected performance of load-balancing policies that do not account for such delays. This is a particularly prominent problem in systems for which the individual units are connected by means of a shared communication medium such as the Internet, ad-hoc networks, wireless LANs. Moreover, the system performance may greatly vary since it incorporates heterogenous nodes that are not necessarily dedicated to the application at hand. In such cases, an actual implementation becomes necessary to understand the load-balancing strategies and their reactions when employed in several environments since mathematical models may not always capture the unpredictable behavior of such systems. In this thesis we propose a software implementation architecture where several distributed load-balancing strategies could be tested and verified under different en-