Merging Parallel Swarms for BitTorrent Performance Improving and Localization
Yang Wang, Jessie Hui Wang, Donghong Qin, Jiahai Yang · 2012
Abstract—With the prevalence of file distribution systems, P2P traffic has caused great challenges to Internet service providers or network operators. The corresponding problems thereby have been a concern for industrial and academic fields. In this paper, we first present a measurement study of torrents and swarms in BitTorrent systems. We find that most swarms suffer from lack of peers and many resource files are shared by multiple trackers. Based on these two observations, we propose a new architecture, named trackers ’ tracker (T-Tracker) to provide a new way of traffic control for network providers, bringing a little change to current BitTorrent protocol. With T-Tracker, our architecture can provide more peers by merging parallel swarms for the performance improvement of BitTor-rent system and provide more choices for biased peer selection of traffic localization schemes. Therefore, quality of BitTorrent service and network utilization can be optimized at the same time. Keywords-BitTorrent; merging swarms; performance improve-ment; P2P traffic localization. I.