Layered communications in load-balancing models

In a multiuser network environment, the computational load varies from one processor to another. A single slow processor can be the cause of bottleneck for the entire application. Load balancing can be used to alleviate the situation and enhance fire performance of the system. This paper presents a variant of the Tinting scheme, Layered communication model, as a means for balancing the workload among the processors. Two new Models: Token passing and Neighboring Queue balancing models are introduced, by applying them to compute the Mandelbrot set. Parallel Virtual Machine (PVM) is used to implement and analyze the performance of various models.