A hybrid load balancing algorithm for coarse-grained applications.

A non-preemptive hybrid load balancing algorithm is proposed for heterogeneous distributed computing environment, since no single load balancing algorithm works well for all kinds of applications and environments. The agents\u27 computing capabilities may also change during runtime because of the background load. This algorithm makes use of the idea of several sub-algorithms. The hybrid model initially classifies the computers and jobs into different groups. Two priority queues are maintained at each worker to record the processes\u27 estimated computing time and the real time. Based on historical experiences, a centralized scheduler can dynamically change the parameters in order to improve the overall performance during runtime. The algorithm balances the work load of coarse-grained applications with interdependent processes such as matrix computation or image processing. Paper copy at Leddy Library: Theses & Major Papers - Basement, West Bldg. / Call Number: Thesis2003 .G35. Source: Masters Abstracts International, Volume: 42-03, page: 0962. Adviser: A. K. Aggarwal. Thesis (M.Sc.)--University of Windsor (Canada), 2003