Mapping the Gnutella Network: Properties of Large-Scale Peer-to-Peer Systems and Implications for System Design Authors: Matei Ripeanu Ian Foster Adriana Iamnitchi
Introduction Gnutella network is an open architecture, self-organizing, and highly scalable P2P system. Quantitative evaluations of P2P systems behaviors are important.
Gnutella Protocol: Design Goals 1.Ability to operate in a dynamic environment 2.Performance and Scalability 3.Reliability 4.Anonymity
Gnutella Network Analysis 1.Growth Trends 2.Estimate of Gnutella Generated Traffic 3.Connectivity in Gnutella Network 4.Mapping between Internet Infrastructure and Gnutella Network
Growth Trends All graphs are from this paper. Memorial Day
Estimate of Gnutella Generated Traffic On average, only 36% of the total traffic (in bytes) is user-generated traffic (Query messages). However, the problem was solved after June 2001 with the newer Gnutella implementation: 92% query messages, 8% PING and others. The estimate traffic for a large Gnutella network is about 1Gbps or 330TB/month: (170,000 connections for 50,000 nodes) X (6Kbps per connection) It amounts to 1.7% of total traffic in US Internet backbones in Dec All graphs are from this paper.
Connectivity in Gnutella Network All graphs are from this paper. Power-Law
Internet Infrastructure and Gnutella Network All graphs are from this paper. Gnutella is completely independent from the Internet structure.
Conclusions Gnutella generates a significant fraction of the total Internet traffic. Node connectivity follows a multi-modal distribution. The topology mismatch between Gnutella and Internet.