A Quantitative Analysis of the Gnutella Network Trac
Demetrios Zeinalipour-Yazti, Theodoros Folias · 2002
Peer-to-Peer (P2P) le-sharing systems such as Gnutella, Morpheus and Freenet have recently attracted a lot of interest from the internet community because they realized a distributed infrastructure for sharing les. Such systems have shifted the Web's Client-Server model paradigm into a Client-Client model. The tremendous success of such systems has proven that purely distributed search systems are feasible and that they may change the way we interact on the Internet. P2P systems uncover many new exciting features such as robustness, scalability and high fault tolerance but with a price. Most research concentrates on optimizing the communication and data model of such systems but inadequate work has been done in area of analyzing such systems. Most approaches tend to use as their basis simulation models which can lead to wrong observations and solutions. In this project we investigate the behavior of the Gnutella system by analyzing large log traces that we have obtained with gnuDC, our Distributed Gnutella Crawler. We describe gnuDC design and implementation choices and we then describe its architecture. We make an analysis of 56 million messages that we obtained with 17 workstations in a 5 hour interval. We have also done an extensive analysis on IP addresses observed in the gnutella network. We believe that our study will facilitate the design of new more ecien t communication algorithms between peers.