Peer-assisted systems that use BitTorrent-like peer-to-peer protocols can potentially support on-demand streaming of popular media files to large numbers of clients. In these systems, the policy used for exchanging content pieces among the participating peers is crucial to system performance. In this paper, we present an evaluation of two candidate piece selection policies, namely Portion [1] and In-Order, using an instrumented BitTorrent client. In particular, the piece selection policies are implemented in the BitTorrent Transmission client, and results are reported for experiments conducted using the instrumented client on a controlled networking testbed. We have observed that the portion policy achieves a good tradeoff between playback startup delay performance and piece diversity.
[1]
Carey L. Williamson,et al.
Analysis of bittorrent-like protocols for on-demand stored media streaming
,
2008,
SIGMETRICS '08.
[2]
B. Cohen,et al.
Incentives Build Robustness in Bit-Torrent
,
2003
.
[3]
Niklas Carlsson,et al.
Peer-assisted On-demand Video Streaming with Selfish Peers
,
2009,
Networking.
[4]
Niklas Carlsson,et al.
Peer-Assisted On-Demand Streaming of Stored Media Using BitTorrent-Like Protocols
,
2007,
Networking.
[5]
Walid Dabbous,et al.
Can Realistic BitTorrent Experiments Be Performed on Clusters?
,
2010,
2010 IEEE Tenth International Conference on Peer-to-Peer Computing (P2P).