On Efficient Data Management In Peer-to-Peer System

Peer-to-Peer (P2P) is a lean, mean distributed computing technology that avoids the delays caused by overloaded servers, dodges hackers, and offers a low-cost alternative to adding expensive servers to overloaded systems. The wide spread use of P2P applications, has attracted much attention by academicians, researchers and practitioners who are working round the clock to resolve some challenges associated with this disruptive technology. One of the problems that large scale P2P applications face is effective data management. Individual machines, which indeed provide storage of resources, are constantly dropping out, or failing, or joining the system. The complexity of the network increases as more machines join.P2P systems are classified into two categories; unstructured and structured. Today, the most popular P2P applications operate on unstructured networks as they are simple, robust and dynamic. Unstructured systems are designed more specifically for the heterogeneous Internet environment, where the nodes' persistence and availability are not guaranteed. Under these conditions, it is impossible to control data placement and to maintain strict constraints on overlay network topology, as structured applications require. Currently, these systems are widely deployed in real life-hence their paramount importance and applicability. Efficient data management is the first step towards realization of an effective P2P system.In this thesis we explore means of improving data management in P2P systems. The main contribution of this work is in two folds. First, we present a modularized data management model that clearly separates the functional components of a P2P system, enabling us to tailor the P2P infrastructure to the specific needs of various Internet applications as well as enhance data management capability for P2P systems. Second, we address data location and routing problem in unstructured P2P systems by introducing a low-cost search method called Intelligent Walks (I-Walks), which applies informed and efficient techniques of data location and outperforms traditional floodbased search schemes including Random Walks.