Modeling and performance analysis of BitTorrent-like peer-to-peer networks
Proceedings of the 2004 conference on Applications, technologies, architectures, and protocols for computer communications
SIGMETRICS '05 Proceedings of the 2005 ACM SIGMETRICS international conference on Measurement and modeling of computer systems
Flash crowd in a file sharing system based on random encounters
Interperf '06 Proceedings from the 2006 workshop on Interdisciplinary systems approach in performance evaluation and design of computer & communications sytems
Analysis of peer-to-peer file dissemination
ACM SIGMETRICS Performance Evaluation Review
On the stability of two-chunk file-sharing systems
Queueing Systems: Theory and Applications
Implications of peer selection strategies by publishers on the performance of P2P swarming systems
ACM SIGMETRICS Performance Evaluation Review
| Hi-index | 0.01 |
We consider a BitTorrent type file sharing algorithm with randomized chunk copying process. The system functions in completely distributed way without any `Tracker', just relying on randomness. In such case the stability becomes an issue. It may happen, say, that some chunk becomes rare. This problem can persist and cause accumulation of peers in the system, resulting in unstable system. The considered algorithms result in processes similar to urn-processes. The rare chunk phenomenon corresponds to Polya-urn type process, where common chunks are favored. However, some urn-processes like the Friedman-urn can provide good balance by favoring rare chunks in copying process. Recently, we showed that an algorithm based on Friedman-urn is efficient in two chunk case. We generalize this algorithm for the more realistic case of many chunks. It shows good performance in terms of balance of chunks in an open system with constant flow of incoming peers. Further, the system is able to cope with instances like `flash crowd', with large burst of incoming peers. The open system can also quickly reach equilibrium after an initial imbalance, when the system starts from a state with one rare chunk. We constructed a simplified model, assuming a good balance of chunks, and get results surprisingly close to simulations for Friedman-urn based random process.