Using Hector to run MPI programs over networked workstations

Networked workstations represent an increasingly popular distributed platfo rm for running large parallel programs. They can present a low-cost alterna tive to purchasing supercomputer time or additional usable computational ca pability, Several capabilities are desirable in order to harness workstatio ns, including support for a widely accepted parallel programming environmen t, task migration, intelligent resource allocation, fault tolerance, and to tally transparent support of these features. The Hector system is designed to provide these capabilities to MPI programs. The structure of the system and experiences using the system on loaded workstations to run scientific c odes are described. Copyright (C) 1999 John Wiley & Sons, Ltd.