Abstract: In this paper，the Gnutella network is basically taken as an object，the essential characteristics and pivotal protocol are analyzed，and a Gnutella content crawler is designed using Limewire source code.The authors measure the properties of file sharing type，file sharing number，file sharing space，file replica，peer daily online time and peer application layer bandwidth in detail，and carry out statistical analysis for these measured properties.The statistical analysis results show that there is significant correlation between file sharing number and sharing space and there is weak correlation between application layer bandwidth and peer daily online time.Furthermore，a cluster analysis is done according to file sharing number，file sharing space，application layer bandwidth and daily online time.The result shows that the common peers is up to 97 percent of the statistical peers in Gnutella network.

Key words: Gnutella, Peer-to-Peer networks, file sharing, cluster analysis

摘要： 以Gnutella网络为测量对象，分析了Gnutella网络的基本特点和关键协议，利用Limewire的源代码设计了一个Gnutella网络内容爬虫，测量了Gnutella网络中用户的共享文件类型、共享文件数目、共享空间、文件副本、用户日均在线时长和节点应用层带宽，并对以上测量结果进行统计分析。统计分析表明文件共享数目与共享空间存在显著的相关关系，应用层带宽与日均在线时长则存在弱相关关系。此外，对Gnutella网络中的用户按照共享文件数目、共享空间、应用层带宽、日均在线时长这4个方面进行了聚类分析，分析结果表明，Gnutella网络中的平庸用户占所统计用户数的97%。

关键词: Gnutella网络, 对等网络, 文件共享, 聚类分析