Autonomic resource discovery service infrastructure for Grid (ARDIG)

Abstract

Traditional Grid monitoring and discovery services are mainly responsible to discover shared computing resources spanning multiple administrative domains and present their unified functionality to the users. Often these discovery services are not fault tolerant, and popular Grid middleware deploy discovery services which do not tolerate high resources churn. Moreover, some Grids deploy centralized discovery service solutions which severely limit the scalability of the Grid. Although solutions such as heart beat systems for dynamic resource availability, super peer discovery services which cater for scalability, have been devised to address this problem. However no unified discovery architecture has been proposed and implemented for next generation dynamic, decentralized, pervasive Grids. In this report, architecture is proposed which integrates an autonomic infrastructure into Grid monitoring and discovery services in order to provide for a more flexible, robust and self-healing Grid. The proposed architecture will lead towards a generic implementation which will be used to transparently extend existing discovery services.