
P2P systems have witnessed phenomenal development in recent years. Simulation has emerged as a dominant tool for evaluating and analyzing new and existing algorithms and techniques in P2P systems. However, most existing simulators are only able to simulate relatively small numbers of nodes. As the size of a typical real world P2P network is in the order of millions of nodes and growing, the demand for a large-scale simulator becomes even more apparent. Distributed simulation can offer a solution to this challenge.
dPeerSim is a distributed simulator for P2P systems, based on an existing sequential simulator, PeerSim. dPeerSim is scalable, only limited by the amount of computational resource available (memory and CPU). The advantage of such a scalable simulator is twofold: firstly, the ability to test systems on networks of real world size. Secondly, with a scalable system, even large simulations can be executed in a reasonable time scale.
dPeerSim is based on the Logical Process paradigm and uses a conservative synchronisation approach. It has been tested and evaluated for different standard structured P2P networks and has been successfully deployed to support our own research on self-organising maintenance mechanisms for very large and realistic P2P networks under churn.
Sources available upon request.
Page
maintained by G. Theodoropoulos. Last update:
03/05/2009