A Measurement Study of Piece Population in BitTorrent

BitTorrent is the most popular peer-to-peer software for file sharing, which has contributed to a significant portion of today´s Internet traffic. Many measurement studies have been devoted to the BitTorrent system at the peer-level; yet few have examined the microscopic piece-level, in particular, the piece populations. This information is very useful in understanding the dynamics and evolution of BitTorrent swarms, and especially the effectiveness of its rarest-first policy that strives to ensure an even distribution of pieces. In this paper, we present a systematic measurement study on the distribution and evolution of the piece population in BitTorrent. Our measurement is based on real BitTorrent data gathered from both the Internet and controlled PlanetLab swarms. The data is collected by multiple administrated clients distributed in different parts of the network, which collectively offer a global view of the piece distribution. We analyze both snapshot data of the near-instantaneous population of pieces in BitTorrent swarms, and long-term data of the evolution of the piece population over several days, especially during the early phases of the swarm´s lifetime. Our results validate that the downloading policy of BitTorrent is quite effective from a piece distribution and evolution perspective; yet enhancements are still possible to achieve the ideal piece distribution.