Design And Building Of The Data-collecting Platform And Study Of The P2P Network

A common method to perform a network research is to study the network snapshots, these can reflect most aspects of the network, this characteristics are important to map a network. We can also get the dynamics by analyzing some sequential snapshots; a platform is needed to crawl these snapshots.In this paper, we introduce a platform that is used for the Gnutella network. The basic theory is ping/pong message in Gnutella protocol, and the hardware is PC; the software is a modification of the Gtk-gnutella, which runs under linux; the database is mysql. We focus on these three aspects to modify the platform and study the factors that influence the performance. We can crawler 15,000 pongs and 300 peers each computer in 1 minute, and can get the information of the connection, file and nodes at the same time. After some cooperation with the other similar platforms, we can get that the bottleneck of a crawling speed is the time used to form a TCP connection.We do some analysis on the network characteristic using our 15-hour data. We get some information about the file type, popularity and network topology. At last we introduce the pollution in P2P network. We give another standard to justify pollution. After some experiments, we include that there is not large-scale pollution in P2P networks.