TU Delft
 
Alexandru IOSUP
P2P peer level view
Parallel and Distributed Systems
EWI PDS A.Iosup Research P2P peer level view
 
Index
 
News
 
Resume
 
Personal Corner
 
Personal Projects
 
Research
 
Industry
 
Teaching
 
Links

Quick links
rationale intro results research publications references

Hairy World -- BitTorrent users per city map of the world Hairy World -- BitTorrent users per city map of the world

Rationale
why and how is this work relevant?
Peer-to-Peer systems like BitTorrent have recently attracted the interest of the Internet audience with their ability to share content at high speeds, while lowering the burden on the initial data owner. Besides optimizing the transfers, researchers are focusing on robustness and fault tolerance, issues inextricably linked with the location and activity of the network users. In this study we present a peer-level analysis of a large set of BitTorrent measurements and the correlation of three views on the location, time patterns and users activity. We present evidence that BitTorrent content presents non-trivial locality features and show that exploiting these features could greatly benefit the world of P2P systems, and in particular BitTorrent.



A Personal Introduction...
how did I get into this?
I was attending the ASCI A9 course and was due to make a small report. The subject was supposed to be in the area of P2P systems -- distributed systems in which nodes (which normally play equal roles, being therefore called peers) employ distributed resources to perform a critical function in a decentralized fashion [6] -- and, since I was interested in the P2P file-sharing systems, and was given the opportunity and much needed assistance, I've started working on the analysis of a large BitTorrent trace, aka The Delft Bittorrent Dataset. The trace and much help have been kindly provided by Drd.Pawel Garbacki and Dr.Johan Pouwelse. Many thanks!



Some results
mostly pics and graphs, taken from the publications...
The Hairy World bubbles movie, presented on February 25, 2004, at the IPTPS'05 workshop.

The movie has ~170MB.

Here's a sample:
The Hairy World bubble-based movie

Some pics taken from the article
Analyzing BitTorrent: Three Lessons from One Peer-Level View

JPG, 300dpi [107.31KB]
Figure 1. The distribution of users per continent: (a) for group Varied; (b) for group Detailed. 		JPG, 300dpi [140.00KB]
Figure 2. The distribution of users per continent: (a) for Applications; (b) for Games.
JPG, 300dpi [126.78KB]
Figure 3. The distribution of users per country: (a) for group Varied; (b) for group Detailed. 		JPG, 300dpi [110.95KB]
Figure 4. The distribution of users per country for category Movies.
JPG, 300dpi [113.51KB]
Figure 5. The distribution of users per country for category Games. 		JPG, 300dpi [118.25KB]
Figure 6. The distribution of users per organization for the Popular Movie subcategories English(EN), German(DE), and French(FR).
JPG, 300dpi [158.67KB]
Figure 7. Time patterns at the country level for group Varied. 		JPG, 300dpi [134.98KB]
Figure 8. The normalized number of users, seeds, and number of file chunks for group Detailed.
JPG, 300dpi [97.38KB]
Figure 9. The number of superpeers and collector peers. 		JPG, 300dpi [314.87KB]
Figure 10. The distribution of the number of points per user.



Research
what did we do, after all?
Three important lessons surface from our study. The most important lesson is that BitTorrent shared contents displays various levels of locality, both trivial and non-trivial. The most obvious types of locality are the communities of users speaking the same language sharing localized versions of some data. The non-trivial examples include the preference of some countries for particular categories of contents (e.g. Hong Kong downloading a soccer management simulator, possibly because of the growing soccer gambling addiction [2]) and the presence of single ISPs that serve over 50% of some media's users. The second lesson is that BitTorrent users are strongly influenced by the presence of alias media---the same contents presented under different names and/or languages. The final lesson is that global views over P2P networks do not yield conclusive characterizations of the network under survey; instead, detailed views organized per categories of files and alias media should be used.

In addition to the three lessons, this study shows that BitTorrent exhibits a number of distinctive features. First, Europe, and not North America, is the most important contributor to BitTorrent's traffic. Second, the percentage of users per continent, country, or organization (from the total number of observed users) is very close to the percentage of their weight (from the total observed weight). Third, the highest number of users can be actually seen during the during Europe's slow work hours, and not only during diurnal and nocturnal hours. Fourth, the total number of chunks available in the BitTorrent is sometimes very low, usually due to web sites failures. Fifth, the probability of a node having or downloading a certain number of files decreases supra-exponentially with the number of files. Sixth, users starting to download a number of files usually finish downloading that number of files or the equivalent. Finally, the majority of the BitTorrent users drop just before obtaining at least one complete file.



Publications, conferences, talks
validating our work...
A.Iosup,P.Garbacki,J.A.Pouwelse,D.H.J.Epema, Analyzing BitTorrent: Three Lessons from One Peer-Level View, ASCI'2005, 6-8 June, Heijen, The Netherlands (accepted).
article Article, PDF [330KB]
 
A.Iosup, Where in the world is Carmen BitDiego? And who is she, anyways..., ASCI A9 Report, February 2005.
report Article, PDF [140KB]
 
A.Iosup, Where in the world is Carmen BitDiego? And who is she, anyways..., Presented at The 12th annual ASCI Computing Workshop (GNARP) 2005, February 2005, Garderen, The Netherlands.
presentation Presentation, PPT [7MB]
venue GNARP Workshop 2005.  



References
these studies have enabled us to work on this project
  1. K. P. Gummadi, R. J. Dunn, S. Saroiu, S. D. Gribble, H. M. Levy, and J. Zahorjan. Measurement, modeling, and analysis of a peer-to-peer file-sharing workload. In SOSP ’03: Proceedings of the nineteenth ACM symposium on Operating systems principles, pages 314–329. ACM Press, 2003.
  2. Hong Kong Govt. Gambling review: A consultation paper, Jul 2001. http://www.info.gov.hk/archive/consult/2001/gambling-e.pdf.
  3. M. Izal, G. Urvoy-Keller, E. Biersack, P. Felber, A. Al Hamra, and L. Garces-Erice. Dissecting BitTorrent: Five months in a torrent’s lifetime. In Passive and Active Measurements (PAM 2004), April 2004.
  4. T. Karagiannis, A. Broido, N. Brownlee, k. claffy, and M. Faloutsos. Is P2P dying or just hiding?. In Global Internet and Next Generation Networks (Globecom 2004), Dallas, Texas, US, Dec 2004.
  5. N. Leibowitz, M. Ripeanu, and A.Wierzbicki. Deconstructing the Kazaa Network. In Proceedings of The Third IEEE Workshop on Internet Applications, page 112, June 23-24 2003. San Jose, California.
  6. X. Li and J. Wu, Searching Techniques in Peer-to-Peer Networks, to appear in Handbook of Theoretical and Algorithmic Aspects of Ad Hoc, Sensor, and Peer-to-Peer Networks, J. Wu (ed.), CRC Press, 2005.
  7. A. Parker. The true picture of peer-to-peer file-sharing. CacheLogic Presentation, July 2004.
  8. J. Pouwelse, P. Garbacki, D. Epema, and H. Sips. The Bit-Torrent p2p file-sharing system: Measurements and analysis. In Proceedings of the 4th International Workshop on Peer-To-Peer Systems (IPTPS’05), Ithaca, New York, USA, February 2005.
  9. S. Saroiu, P. Gummadi, and S. Gribble. A measurement study of peer-to-peer file sharing systems. In Proceedings of Multimedia Computing and Networking, 2002.
  10. R. Schollmeier and G. Kunzmann. GnuViz - Mapping the Gnutella Networks to its Geographical Locations. Praxis der Informationsverarbeitung und Kommunikation (PIK), 26(2):74–79, 2003.
  11. S. Sen and J. Wong. Analyzing peer-to-peer traffic across large networks. In Second Annual ACM Internet Measurement Workshop, November 2002.
  12. D. Zeinalipour-Yazti and T. Folias. A quantitative analysis of the gnutella network traffic. Course Project for Advanced Topics in Networks, with M. Faloutsos at the University of California - Riverside, Dpt. of CS, April 2002.



                                                                                                                                                                                                                                             
New! Feedback


Your feedback is important to this research! Please do not hesitate to drop a line, comment, raise questions, or point us to new directions. Thank you!

Your name
Your email address
Message subject
Message contents



     

Top of the page

Last modified: Mon, 22 August, 2005 4:06 PM
The newest version of this page can be found at: http://www.pds.ewi.tudelft.nl/~iosup/p2pview.html
Copyright © 1998-2005 Alexandru Iosup. All Rights Reserved.
Google Analytics .