Performance analysis of bittorrent-like systems with heterogeneous users

Abstract—BitTorrent is the most successful peer-to-peer sys-tem and has attracted a lot of attention from the research community. Researchers have studied a number of aspects of the system, including its scalability, performance, efficiency and fairness. However, the complexity of the system has forced most researchers to make a number of simplifying assumptions, e.g. user homogeneity, or even ignore some aspects of the protocol altogether, e.g. the Tit-for-Tat (TFT) unchoking scheme, in order to keep the analysis tractable. Motivated by this, in this paper we propose two analytical models that accurately predict the performance of the system without compromising on the realism of the modeling methodol-ogy. Our first model is a steady-state one, in the sense that it is valid during periods of time where the number of users remains fixed. Freed by the complications of time-dynamics, we account for all the details of the BitTorrent protocol, including TFT and optimistic unchoking, and predict a number of performance metrics including upload and download rates, as well as file download delays. Our second model builds upon prior work on fluid models for BitTorrent. Using our first model, we extend the fluid-based methodology to capture the transient behavior as new users join or old users leave, while fully modelling central properties of the BitTorrent system, e.g. TFT. Finally, as an example of how to use our analytical models to study variations of the basic BitTorrent scheme and make design decisions, we propose a flexible token-based scheme for BitTorrent that can be used to tradeoff between overall system performance and fairness, and evaluate the scheme’s parameters that achieve a target operational point. Index Terms—Heterogeneous P2P networks, BitTorrent, per-formance analysis, token based scheme, fairness/delay tradeoff