1 / 27

Takashi Koizumi , Masahiro Yoshida, Satoshi Ohzahata, Konosuke Kawashima

An Analysis of the P2P Traffic Characteristics on File Transfers Between Prefectures and Between Autonomous Systems in the Winny Network. 2 ). 1). Takashi Koizumi , Masahiro Yoshida, Satoshi Ohzahata, Konosuke Kawashima Tokyo University of Agriculture and Technology,

laurie
Download Presentation

Takashi Koizumi , Masahiro Yoshida, Satoshi Ohzahata, Konosuke Kawashima

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. An Analysis of the P2P Traffic Characteristics on File Transfers Between Prefectures and Between Autonomous Systems in the Winny Network 2) 1) Takashi Koizumi, Masahiro Yoshida, Satoshi Ohzahata, Konosuke Kawashima • Tokyo University of Agriculture and Technology, • The University of Tokyo, • The University of Electro-Communications, 3) 1) 1) 2) 3) APCC2010

  2. Outline • Introduction • File distribution protocol in Winny • Analysis method • Definition of the file uploading peers • Definition of the file transfer between peers • Measurement environment • Measurement results • Analysis between prefectures • Analysis between ASs • Conclusion APCC2010

  3. Outline • Introduction • File distribution protocol in Winny • Analysis method • Definition of the file uploading peers • Definition of the file transfer between peers • Measurement environment • Measurement results • Analysis between prefectures • Analysis between ASs • Conclusion APCC2010

  4. Introduction • P2P file sharing networkgenerates huge traffic. • it has an enormous impact to backbone network. • however, the technology of P2P is useful. • VOD, Video streaming, etc... ⇒Analyzing P2P traffic is important. • Problems: • many ISPs do not make public reporton their traffic data. • difficulty of identifying file transfers • from where did peer download files? P2P network APCC2010

  5. Introduction • Analysis methodology • identify the file transfers between peers in Winny network. • Analyze the characteristics of the traffic matrix. • focus on between ASs and between Japanese Prefectures. • Winny network • one of the most popular P2P file sharing applications in Japan. • it generates 20% of traffic in one ISP in 2008. • Reveal the factor of increasing traffic from P2P file-sharing. APCC2010

  6. Outline • Introduction • File distribution protocol in Winny • Analysis method • Definition of the file uploading peers • Definition of the file transfer between peers • Measurement environment • Measurement results • analysis between prefectures • analysis between ASs • Conclusion APCC2010

  7. File distribution protocol in Winny • Feature of Winny network • unstructured P2P network. • layered structure. • diffuse metadata of files (KEY) • 2 types of query. File uploading peer diffuses KEYsto other peers. File searching peer generates search queries to find KEYs. Download the file from the location of file uploading peer included in the KEY. File uploading peer File searching peer APCC2010

  8. Outline • Introduction • File distribution protocol in Winny • Analysis method • Definition of the file uploading peers • Definition of the file transfer between peers • Measurement environment • Measurement results • analysis between prefectures • analysis between ASs • Conclusion APCC2010

  9. Definition of the file uploading peers • Analyze time series of file KEY. • focus on TTL and the file location in the KEY. • 1500 < TTL < 2000. • file location is same with the KEY having peer. Crawler : File Peer A search query : File KEY query hit 1500 < TTL < 2000 file location is Peer A. Peer A is file uploading peer • ※TTL should be smaller than 2000 seconds to exclude spam KEYs. • TTL of the KEY in file uploading peer is set to be greater than 1500 seconds. APCC2010

  10. Outline • Introduction • File distribution protocol in Winny • Analysis method • Definition of the file uploading peers • Definition of the file transfer between peers • Measurement environment • Measurement results • analysis between prefectures • analysis between ASs • Conclusion APCC2010

  11. Definition of the file transfer between peers • process of the file transfer. • changes of a KEY having peer to a file uploading peer. ① Diffusion of KEY Peer B Peer A ② Download demand ② Peer Bis the KEY having peer ① Peer Ais the file uploading peer with continuous crawling ③ Peer B becomes a file uploading peer Peer B Peer A : File : File KEY ④ file transfer to Peer B from Peer A APCC2010

  12. Outline • Introduction • File distribution protocol in Winny • Analysis method • Definition of the file uploading peers • Definition of the file transfer between peers • Measurement environment • Measurement results • analysis between prefectures • analysis between ASs • Conclusion APCC2010

  13. Measurement environment 6 crawlers • Winny Crawling System collects KEYs. • crawl more than 100,000 peers within 10 minutes. • Measurement period • Aug. 4, 2009 01:00 ~(12 hours) Management server DB server APCC2010

  14. Measurement environment 6 crawlers • Types of measured file. • 15 popular files (selected in 1 hour crawling). • books, movies and compressed files. • average file size is 245MB Management server DB server APCC2010

  15. Outline • Introduction • File distribution protocol in Winny • Analysis method • Definition of the file uploading peers • Definition of the file transfer between peers • Measurement environment • Measurement results • analysis between prefectures • analysis between ASs • Conclusion APCC2010

  16. Measurement results • Analysis of the file transfer characteristics • focus on between prefectures and between ASs. ⇒Origin-Destination file transfer matrix. • measurement summary • about 220,000 peers, 280 ASs. • about 30,000 instances of the file transfers between 33,000 peers ※ results are for 47 prefectures and top 50 ASswith number of file transfersabout 15 popular files. APCC2010

  17. Outline • Introduction • File distribution protocol in Winny • Analysis method • Definition of the file uploading peers • Definition of the file transfer between peers • Measurement environment • Measurement results • analysis between prefectures • analysis between ASs • Conclusion APCC2010

  18. File transfer traffic matrix between prefectures • src prefectures on Y-axis, dst prefectures on X-axis • large portion of file transfers are generated in few prefectures. % low high Destination Prefectures Tokyo(src) Source Prefectures Osaka(src) Tokyo(dst) Osaka(dst) APCC2010

  19. File transfer traffic matrix between prefectures • the sum of elements is 100% for each row. • similar traffic distribution in every prefectures. low high Destination Prefectures Source Prefectures within-prefectures src and dst is same prefecture APCC2010

  20. Comparison of Winny and telephone traffic • upper: Winny • below: telephone traffic (in 2007) • traffic within-prefectures • Winny: 2% • telephone traffic: 70% • similar prefectures are selected as destinations. • Tokyo, Osaka, etc. Tokyo Osaka APCC2010

  21. Comparison of Winny and Internet • upper: Winny • below: Internet (in 2008)* • traffic within-prefectures. • Winny: 2% • Internet: 3% • few locality of traffic. • distribution patterns are similar. * Kenjiro Cho, Kensuke Fukuda, Hiroshi Esaki, Akira Kato, Jun Murai, “Where does all the traffic go? – observing trends in Japanese residential traffic –,” International Workshop on Protocols for Future, Large-Scale and Diverse Network Transports, 2009. APCC2010

  22. Outline • Introduction • File distribution protocol in Winny • Analysis method • Definition of the file uploading peers • Definition of the file transfer between peers • Measurement environment • Measurement results • analysis between prefectures • analysis between ASs • Conclusion APCC2010

  23. File transfer traffic matrix between ASs • src ASs on Y-axis, dst ASs on X-axis • large portion of file transfers is generated by few ASs. • 20 ASs generated 85% of total number of file transfers. % low high Destination ASs Source ASs APCC2010

  24. File transfer traffic matrix between ASs • the sum of elements is 100% for each row. • traffic distribution is similar in each AS. • very poor locality of AS network in Winny. low high Destination ASs Source ASs within-AS APCC2010

  25. Outline • Introduction • File distribution protocol in Winny • Analysis method • Definition of the file uploading peers • Definition of the file transfer between peers • Measurement environment • Measurement results • analysis between prefectures • analysis between ASs • Conclusion APCC2010

  26. Conclusion • Analyzed traffic characteristics in Winny network. • many file transfers are generated by few prefectures or ASs. • since there is no mechanism to consider IP network, many file transfers are not within-prefecture or within-ASs. • Future work • long term measurement. • analysis and measurement of other P2P networks (BitTorrent, Share, etc.). APCC2010

  27. thank you very much for your kind attention. APCC2010

More Related