Abstract:Through the study of different computing system architecture, a load balancing strategy with fault-tolerence and adaptive parallelism is designed for desktop grid master-worker programming model. Using Java technology, prototype systems with this load balancing strategy has been setup on campus network. Different complexity Monte Carlo simulation applications as benchmark, ideal speedup is achieved. The total computing capacity which is used exceeds 90%.

Adaptive Parallelism Load Balancing Strategy Design and Realize for Desktop Grid