Cluster computing: The wave of the future?

Abstract

Using clusters of workstations is becoming a very popular method to solve both large and small scientific problems. One of the driving forces behind this shift is the availability of portable robust software to utilize and manage a cluster of workstations.

PVM (Parallel Virtual Machine) is a byproduct of the heterogeneous network research project going on at Oak Ridge National Laboratory and University of Tennessee. It is a software package that permits a user defined collection of serial, parallel, and vector computers appear as one large distributed memory computer. PVM's popularity is quickly making it the worldwide de facto standard for distributed computing.

This talk will describe the driving forces behind cluster computing and the limitations imposed by this method of computing. It will describe the wide breadth of applications being developed in PVM and the features of the latest release of PVM. The talk will also raise several research and social issues that must be addressed before we can make the next big leap in distributed computing.

This work was supported in part by the Applied Mathematical Sciences subprogram of the Office of Energy Research, U.S. Department of Energy, under Contract DE-AC05-84OR21400