Mapping the Gnutella Network: Properties of Large-Scale Peer-to-Peer Systems and Implications for System Design

By Matei Ripeanu, Ian T Foster, Adriana Iamnitchi. arXiv.org, 2002

985 citations recorded by the source corpus.

Despite recent excitement generated by the peer-to-peer (P2P) paradigm and the surprisingly rapid deployment of some P2P applications, there are few quantitative evaluations of P2P systems behavior. The open architecture, achieved scale, and self-organizing structure of the Gnutella network make it an interesting P2P architecture to study. Like most other P2P applications, Gnutella builds, at the application level, a virtual network with its own routing mechanisms. The topology of this virtual network and the routing mechanisms used have a significant influence on application properties such as performance, reliability, and scalability. We have built a "crawler" to extract the topology of Gnutella's application level network. In this paper we analyze the topology graph and evaluate generated network traffic. Our two major findings are that: (1) although Gnutella is not a pure power-law network, its current configuration has the benefits and drawbacks of a power-law structure, and (2) the Gnutella virtual network topology does not match well the underlying Internet topology, hence leading to ineffective use of the physical networking infrastructure. These findings guide us to propose changes to the Gnutella protocol and implementations that may bring significant performance and scalability improvements. We believe that our findings as well as our measurement and analysis techniques have broad applicability to P2P systems and provide unique insights into P2P system design tradeoffs.

Subfields: Computer Networks and Communications, Statistical and Nonlinear Physics

View Paper: https://www.semanticscholar.org/paper/7ac5671bc416cebed36caf5f0e825655e23f8f3c

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Networks and Communications

In Statistical and Nonlinear Physics

Citations in the corpus

Evolution and Structure of the Internet: Index (2004, methodology): “The precise distribution is also changing in time, and the most recent measurements (Ripeanu et al., 2002) show two separate regimes at the small and large degree values.”
On Death, Taxes, and the Convergence of Peer-to-Peer and Grid Computing (2003, background): “First-and second-generation P2P collaborations have been characterized at the level of both individual nodes (behavior, resources [2, 46]) and network properties (topological properties [42], scale [32], traffic [47]), revealing not only general resilience but also unexpected emergent properties.”
Community Structure in Large Networks: Natural Cluster Sizes and the Absence of Large Well-Defined Clusters (2008, neutral): “75 Gnutella P2P network on March 30 2000 [143] Gnutella-31 62,561 147,878 0.”
SybilInfer: Detecting Sybil Nodes using Social Networks (2009, background): “Social networks exhibit a scale-free (or power law) node degree topology [21].”
Epidemic spreading in real networks: an eigenvalue viewpoint (2003, background): “However, there is overwhelming evidence that real networks (including social networks [21], router and AS networks [6], and Gnutella overlay graphs [ 22 ]) deviate from such homogeneity|they follow a power law structure instead.”
Giggle: A Framework for Constructing Scalable Replica Location Services (2002, background): “Gnutella [26] does not use indexing mechanisms; its relatively good search performance (as measured in number of hops) is due to intensive network usage.”
Statistical properties of community structure in large social and information networks (2008, neutral): “Gnutella 62,561 147,878 P2P network [29]”
Controlling edge dynamics in complex networks (2011, neutral): “p2p-2 8,846 31,839 [28,56] A sent messages to B 30.”
A Taxonomy of Botnet Structures (2007, background): “Large numbers of connections on a client host are highly unusual, even for P2P software [33, 49].”
MAAN: A Multi-Attribute Addressable Network for Grid Information Services (2003, background): “However, this approach does not scale well because of the large volume of query messages generated by flooding [13, 17].”
PocketLens: Toward a personal recommender system (2004, methodology): “We call this the random discovery architecture because the topology of the client connections has been shown to be similar to a random graph [Ripeanu et al. 2002; Jovanovic 2001], and the find neighbor module incorporates neighbors into the model without evaluating them first.”