PARALLEL AND DISTRIBUTED ALGORITHMS: PRACTICAL APPROACHES

The book is based on two representative parallel models namely: single program multiple data (SPMD) with fault tolerance and multiple programs multiple data (MPMD) with fault tolerance, which are used to evaluate the performance of parallel and distributed algorithms. These two models have been mapped to a cluster of workstations considering concurrent tasks with a goal to achieve better performance and cost effectiveness of parallel and distributed applications. Upon completion of each task, the results are collected by the master computer to produce the final results within the bounded execution time. The performance parameters such as serial time Sr(p), parallel time T(p), percentage (%) of execution time saving, through-put Th(p), speed-up S(p), Efficiency E(p), redundancy R(p), utilization U(p, and quality Q(p) are used for evaluating the performance of parallel and distributed algorithms, using N number of workstations in a cluster environment. In fact, workstations are interconnected with Ethernet LAN of 10 mbps data rate or more.