Distributed P2p Simulation Technology

P2P (Peer-to-Peer) has become one of the hottest issues in information technology domain recently, which has attracted the attention both of industry and academia. Since the P2P network is a complex, large scale and high dynamic system, it is prohibitively expensive and unpractical to research P2P technology by deploying it in an existing network environment. A powerful P2P simulator plays an increasing important role in evaluating the new or existing P2P applications. For enlarging the simulation scale and improving the simulation fidelity, this paper researches the P2P simulation technology based on the distributed simulation.At first, this paper surveys the related work of P2P simulation, describes the designe principle of distributed P2P simulator and the key techniques to resolve, and designes the distributed P2P simulation environment. Based on the above work, the main contribution of this paper is proposing the architecture of Double Engines based Distributed P2P Simulator (DEDPS) and Degree-sequence and Bandwidth-first based Partition Algorithm (DBPA).The architecture of DEDPS is composed of peer simulation engine which is controlling the function of simulating peer and network simulation engine which is taking charge of the network communication. Under this architecture, the Simulation Runtime Infrastructure (SRI) and the Common Basic Infrastructure (CBI) is separated, which makes the DEDPS have good extensibility. In DEDPS, the peer simulation engine can take advantage of simulation application computer to support a small scale P2P network simulation. Assisted by the network simulation engine, the peer simulation engine can transfer the simulation data as network packet to other simulation application computers or real P2P applications, which not only can realize large scale P2P network simulation, but also can improve simulation fidelity. Meanwhile, this paper researches and analyzes the key techniques of peer simulation engine and network simulation engine, and proposes the standards and technical resolving methods.DBPA is mainly applied for distributed P2P simulation load balance on TFSL (Topology First, Simulation Last) simulation model. This paper researches and analyzes the patition algorithm by various experiments of random network topology partition and power-law network topology partition, and verifies the capability, efficiency and feasibility of DBPA with the analysis of load balance, edge-cut balance, edge-cut rate and time complexity, while compares with other partition algorithms.At last, this paper designes and realizes a DEDPS based on the architecture proposed and some related key techniques, and testes the DEDPS by Gnutella protocol simulation experiments from the aspect of simulation scale and simulation fidelity. With the analysis of simulation results, it shows the DEDPS can execute large scale P2P simulation and improve simulation fidelity.