Building Scalable Grid Resource Management Systems Based on the P2P Paradigm

Abstract

During the past decade, Grid computing has gained a lot of interest in the computing research community. A Grid is a geographically distributed computation platform comprising a set of heterogeneous resources (e.g. computers, data sources, applications, etc.) that users can efficiently access and integrate through a single interface. Grid computing differs from conventional distributed computing in that it focuses on large-scale resource sharing, offers innovative applications, and, in some cases, is geared towards high-performance systems. The main subsystem responsible for enabling such integration is the Grid Resource Management System (GRMS). However, in order to facilitate resource sharing, it would first be necessary that the GRMS finds the location of desired I resources in an efficient and timely fashion, a problem known as resource discovery. Current approaches to resource discovery on Grids are based on centralized or hierarchical client/server models. As Grid systems increase in scale, they begin to require solutions for issues of self-configuration, fault­ tolerance, and scalability, for which peer-to-peer (P2P) research has much to offer.

Augmenting the peer-to-peer model to the services of the Grid promises to eliminate bottlenecks and ensure greater scalability, availability, and fault­ tolerance.