Modeling file popularity in peer-to-peer file sharing systems

From the birth of Peer-to-Peer (P2P) protocols, Internet traffic has changed its characteristics; the amount of traffic carried by the web has grown exponentially with strong effects on traffic matrixes, both for the distributed nature of P2P protocols and for their intense bandwitdh request. Many research works studied the features and the behaviors of such P2P protocols, simulating the overlay network topologies at the application layer. We present a scalable P2P model, which is particularly focused on file popularity, considered as the level of interest that a file arouses in the system. The work has the aim of describing the effects that such a quantity generates in the query behavior, assuming that it is strongly linked to marketing procedures. For this reason we make use of the product life cycle in marketing environment to model such level of interest. The model can represent huge overlay networks, up to 500000 peers. The properties of such a model are studied considering the resulting file query distribution, by running multiple independent replications for every simulation. The number of queries performed by every peer and the peer' uptime are also considered, in order to validate the model, comparing simulation output with the measurement results in Gnutella network.