Efficient Normal Peers Group Recovery in Hierarchical Peer-To-Peer

Abstract

While structured peer-to-peer (P2P) offers benefits on its scalability and efficiency in performing a successful query lookup, its stability is suffered from the dynamics of the overlay structure caused by churn. In a structured hierarchical P2P, in addition to potential decreases of the system performance, the superpeer failure cases also forced the normal peers under its responsibility to disconnect from the system. The ultimate goal of our work is to develop a collective rejoin algorithm that provides an efficient mechanism for the normal peers to rejoin to the system once a superpeer failure occurs. \Ve implemented hvo-layer hierachy architecture, in \Vhich nodes with higher capability arc placed in the upper layer and organized in a Chord ring. These nodes act as a gate\vay for the lower layer in \Vhich other nodes are grouped. Each group is organized in a star structure, and each member of the group is connected directly to the related gate,vay node. We ·expect our proposed architecture and algorithm produces less traffic load than the individual rejoin approach. Thus, the performance degradation caused by the normal peers rejoin process, as an impact of churn, can be minimized.