G2G: A Meta-Grid Framework for the Convergence of P2P and Grids

G2G: A Meta-Grid Framework for the Convergence of P2P and Grids

Wu-Chun Chung (National Tsing Hua University, Taiwan), Chin-Jung Hsu (National Tsing Hua University, Taiwan), Yi-Hsiang Lin (National Tsing Hua University, Taiwan), Kuan-Chou Lai (National Taichung University, Taiwan) and Yeh-Ching Chung (National Tsing Hua University, Taiwan)
DOI: 10.4018/978-1-4666-0056-0.ch017
OnDemand PDF Download:
No Current Special Offers


Grid systems integrate distributed resources to form self-organization and self-management autonomies. With the widespread development of grid systems around the world, grid collaboration for large-scale computing has become a prevalent research topic. In this paper, the authors propose a meta-grid framework, named the Grid-to-Grid (G2G) framework, to harmonize autonomic grids in realizing a grid federation. The G2G framework is a decentralized management framework that is built on top of existing autonomic grid systems. This paper further adopts a super-peer network in a separate layer to coordinate distributed grid systems. A super-peer overlay network is constructed for communication among super-peers, thus enabling collaboration among grid systems. This study proposes the G2G framework for use in a Grid-to-Grid federation and implements a preliminary system as a demonstration. Experimental results show that the proposed meta-grid framework can improve system performance with little overhead.
Chapter Preview

G2g Framework And Prototype

Currently, most grid systems are deployed according to centralized or hierarchical management approaches. However, these approaches have poor performance in terms of scalability, resiliency, and load-balancing for managing distributed resources (Mastroianni et al., 2007). Centralization and hierarchy are the weaknesses of deploying large multi-institutional grid systems, let alone in the widely internetworking system. In general, the performance by adopting the super-peer model is more efficient and convenient than that without adopting the super-peer model in large-scale computing environments.

Complete Chapter List

Search this Book: