LBSim: A simulation system for dynamic load-balancing algorithms for distributed systems.

In a distributed system consisting of autonomous computational units, the total computational power of all the units needs to be utilized efficiently by applying suitable load-balancing policies. For accomplishing the task, a large number of load balancing algorithms have been proposed in the literature. To facilitate the performance study of each of these load-balancing strategies, simulation has been widely used. However comparison of the load balancing algorithms becomes difficult if a different simulator is used for each case. There have been few studies on generalized simulation of load-balancing algorithms in distributed systems. Most of the simulation systems address the experiments for some particular load-balancing algorithms, whereas this thesis aims to study the simulation for a broad range of algorithms. After the characterization of the distributed systems and the extraction of the common components of load-balancing algorithms, a simulation system, called LBSim, has been built. LBSim is a generalized event-driven simulator for studying load-balancing algorithms with coarse-grained applications running on distributed networks of autonomous processing nodes. In order to verify that the simulation model can represent actual systems reasonably well, we have validated LBSim both qualitatively and quantitatively. As a toolkit of simulation, LBSim programming libraries can be reused to implement load-balancing algorithms for the purpose of performance measurement and analysis from different perspectives. As a framework of algorithm simulation can be extended with a moderate effort by following object-oriented methodology, to meet any new requirements that may arise in the future.Dept. of Computer Science. Paper copy at Leddy Library: Theses & Major Papers - Basement, West Bldg. / Call Number: Thesis2004 .D8. Source: Masters Abstracts International, Volume: 43-05, page: 1747. Adviser: A. K. Aggarwal. Thesis (M.Sc.)--University of Windsor (Canada), 2004