Gauging the network friendliness of P2P applications

This demo presents a software tool whose aim is to quantify the level of network awareness and friendliness of the currently deployed Internet P2P applications. Considering such systems as black-boxes, we passively measure the traffic they generate and we gather, by means of both passive inference and active-probing techniques, some quantities relevant for their description. Observed metrics include, e.g., the amount of traffic that the peer under observation exchanges with peers belonging to the same Autonomous System or geographical Country. Similarly, we inspect the traffic exchanged in terms of its proximity in the underlying IP networks, expressed e.g., as RTT delay or number of IP hops. A number of statistics are then computed, in real-time, over the peer database inferred from measurement (e.g., ranging from empirical probability mass function, to correlation among the measured parameters, etc.). A flexible user interface displays a variety of data representations such as pmf, scatter plots, geographical maps and exploits Kiviat charts to represent, at a glance, the full range of measured values in a compact and visually-intuitive way.