LEDBAT+BitTorrent Dataset

In (about) 2010, BitTorrent announced that he would drop TCP in favor of a new experimental congestion control protocol, initially closed source and nicknamed uTP and later standardized at the IETF under the name of LEDBAT. In this project, we study the impact that LEDBAT has on BitTorrent performance with an experimental approach. From a more abstract and fundamental viewpoint, we study the interplay of data plane throughput vs control plane delay. We started with a simulative approach in [P2P-11], then continued with an experimental approach in [TMA-12] and [P2P-13a], gathering datesets that we release in this page.

Aim

In particular project, we address the trade-off between the data plane efficiency and the control plane timeliness for the BitTorrent performance. We argue that loss-based congestion control protocols can fill large buffers, leading to a higher end-to-end delay, unlike low-priority or delay-based congestion control protocols. We perform experiments for both the uTorrent and mainline BitTorrent clients, and we study the impact of uTP (a novel transport protocol proposed by BitTorrent) and several TCP congestion control algorithms (Cubic, New Reno, LP, Vegas and Nice) on the download completion time.

Findings

Briefly, in case peers in the swarm all use the same congestion control algorithm, we observe that the specific algorithm has only a limited impact on the swarm performance. Conversely, when a mix of TCP congestion control algorithms coexists, peers employing a delay-based low-priority algorithm exhibit shorter completion time. To promote further analysis and cross-comparison, we additionally make all our dataset available to the research community.

Dataset

Dataset includes application-level logs, kernel-level queue log and traffic volume statistics gathered in multiple tests executed on the Grid5000 testbed.

• Details on the file/folder structure in each dataset are available in this README
• Details of the application-level log format are also available in this legend

Specifically, for each test we varied the TCP congestion control flavor, or used the BitTorrent version using the uTP congestion control protocol.

• uTP 4.5GB , details in [TMA-12]
• Vegas 0.8GB, details in [P2P-13a]
• Cubic 0.8GB, details in [P2P-13a]
• NewReno 2.7GB, details in [P2P-13a]
• TCP-LP 2.6GB, details in [P2P-13a]

References

• [P2P-11] Testa, C. and Rossi, D., "The impact of uTP on BitTorrent completion time" IEEE Peer to Peer (P2P’11) sep. 2011, Conference Bibtex

  @inproceedings{DR:P2P-11,
    author = {Testa, C. and Rossi, D.},
    title = {The impact of uTP on BitTorrent completion time},
    booktitle = {IEEE Peer to Peer (P2P'11)},
    year = {2011},
    month = sep,
    address = {Kyoto, Japan},
    annote = {category=article state=published project=trans},
    url = {http://www.enst.fr/~drossi/paper/rossi11p2p.pdf},
    howpublished = {http://www.enst.fr/~drossi/paper/rossi11p2p.pdf},
    www_txt_url = {http://www.enst.fr/~drossi/paper/rossi11p2p.pdf},
    note = {keyword=p2p,ledbat}
  }
  
• [TMA-12] Testa, C. and Rossi, D. and Rao, A. and Legout, A., "Experimental Assessment of BitTorrent Completion Time in Heterogeneous TCP/uTP swarms" Traffic Measurement and Analysis (TMA) Workshop at Passive and Active Measurement (PAM) mar. 2012, Conference Abstract Bibtex

  @inproceedings{DR:TMA-12,
    author = {Testa, C. and Rossi, D. and Rao, A. and Legout, A.},
    title = {Experimental Assessment of BitTorrent Completion Time in Heterogeneous TCP/uTP swarms},
    booktitle = {Traffic Measurement and Analysis (TMA) Workshop at Passive and Active Measurement (PAM)},
    year = {2012},
    pages = {52-56},
    address = {Wien, AU},
    month = mar,
    annote = {category=article state=published},
    url = {http://www.telecom-paristech.fr/~drossi/paper/rossi12tma.pdf},
    howpublished = {http://www.telecom-paristech.fr/~drossi/paper/rossi12tma.pdf},
    www_txt_url = {http://www.telecom-paristech.fr/~drossi/paper/rossi12tma.pdf},
    note = {keyword=p2p,bittorrent,ledbat}
  }
  

  BitTorrent, one of the most widespread used P2P application for file- sharing, recently got rid of TCP by introducing an application-level congestion control protocol named uTP. The aim of this new protocol is to efficiently use the available link capacity, while minimizing its interference with the rest of user traffic (e.g., Web, VoIP and gaming) sharing the same access bottleneck. In this paper we perform an experimental study of the impact of uTP on the torrent completion time, the metric that better captures the user experience. We run BitTorrent applications in a flash crowd scenario over a dedicated cluster platform, under both homogeneous and heterogeneous swarm population. Exper- iments show that an all-uTP swarms have shorter torrent download time with respect to all-TCP swarms. Interestingly, at the same time, we observe that even shorter completion times can be achieved under mixtures of TCP and uTP traffic, as in the default BitTorrent settings.
• [P2P-13a] Testa, C. and Rossi, D. and Rao, A. and Legout, A., "Data Plane Throughput vs Control Plane Delay: Experimental Study of BitTorrent Performance " IEEE P2P’XIII sep. 2013, Conference Abstract Bibtex

  @inproceedings{DR:P2P-13a,
    title = {Data Plane Throughput vs Control Plane Delay: Experimental Study of
    BitTorrent Performance },
    author = {Testa, C. and Rossi, D. and Rao, A. and Legout, A.},
    year = {2013},
    booktitle = {IEEE P2P'XIII},
    address = {Trento, Italy},
    month = sep,
    annote = {category=article state=published project=mplane},
    note = {keyword=ledbat,bittorrent},
    url = {http://www.telecom-paristech.fr/~drossi/paper/rossi13p2p-a.pdf},
    howpublished = {http://www.telecom-paristech.fr/~drossi/paper/rossi13p2p-a.pdf},
    www_txt_url = {http://www.telecom-paristech.fr/~drossi/paper/rossi13p2p-a.pdf}
  }
  

  In this paper, we address the trade-off between the data plane efficiency and the control plane timeliness for the BitTorrent performance. We argue that loss-based congestion control protocols can fill large buffers, leading to a higher end-to-end delay, unlike low-priority or delay-based congestion control protocols. We perform experiments for both the uTorrent and mainline BitTorrent clients, and we study the impact of uTP (a novel transport protocol proposed by BitTorrent) and several TCP congestion control algorithms (Cubic, New Reno, LP, Vegas and Nice) on the download completion time. Briefly, in case peers in the swarm all use the same congestion control algorithm, we observe that the specific algorithm has only a limited impact on the swarm performance. Conversely, when a mix of TCP congestion control algorithms coexists, peers employing a delay-based low-priority algorithm exhibit shorter completion time.