Design of an artificial-neural-network-based extended metacomputing directory service

Abstract

This paper analyzes a serious limitation of existing metacomputing directory service of Globus project that the existing metacomputing directory service doesn't support application-oriented queries, and then designs an artificial-neural-network-based GRC (grid resources classifier) to eliminate this limitation. This classifier extends the metacomputing directory service by classifying grid resources into application-oriented categories. The classification precision of this GRC can be continuously improved by self-learning. This kind of new metacomputing directory service will be compatible with the old ones. Thus, the practicability of metacomputing directory service will be improved. © Springer-Verlag 2004.