Torrent Project: Tech Checkpoint 2 (Eric Hartzog)

While implementing a BitTorrent type system into iTunes wouldn’t be difficult, there are a number of seemingly minor issues which can greatly affect the efficiency of the BitTorrent solution.

First of all it is important to establish under what circumstances BitTorrent will outperform traditional client-server direct downloads. While conducting my own testing of the current iTunes video service, I was able to download an 878 MB movie, which is 1:21 worth of medium quality video, in roughly 45 minutes. This translates to an average download speed of around 320 KB/s, and as I observed the download I was able to confirm that the speed of the download was roughly constant for the duration of the download. When downloading movies from current illegitimate BitTorrent providers, the download speeds are normally much slower, averaging around 4 to 6 hours for the same 800 MB download which translates to a 45 KB/s average download rate, which is only 15% of the iTunes download speed.

However, the speed difference can be explained by the scarcity of seeds that have as high an upload bandwidth of the iTunes servers. This means that if iTunes can sustain the proper amount of high-bandwidth seeds for a file, then BitTorrent will be able to have much higher download rates [4]. What my short testing of the current iTunes video download system doesn’t take into account is any type of “flash crowd” which suddenly increase the demand for a specific movie. As I was browsing the iTunes video library, the selection was scarce and none of the movies there would have caused such a flash crowd. Such a flash crowd would cause a dramatic decrease in the download speeds from the iTunes servers while not having nearly as negative effect on the BitTorrent download speeds.

Another important question is Apple’s current implementation of DRM and how it would need to be altered to accommodate a BitTorrent system. As it currently works, iTunes provides a DRM wrapper that is specific to each individual user who downloads content. The benefit of this system is that a user-specific DRM wrapper ensures that only the user who it belongs to can get the correct key from the iTunes servers to decrypt the content. This lowers the chance that a program like DeCSS can be created which can crack the DRM on all the content protected by the iTunes DRM. A legitimate DRM system encourages content providers to license their movies and shows to be downloaded, and if the system is easily cracked then they will withhold licenses.

This is a problem since the BitTorrent solution requires a reworking of iTunes’s current DRM, since all users in a BitTorrent swarm must be working with the exact same copy of a file. Although there can still be a general DRM wrapper around video, no longer can user-specific DRM prevent a single program from removing the DRM from any downloaded file. While this doesn’t eliminate the DRM, it potentially weakens it and may cause content providers to be less likely to give out the licenses necessary for the video service to gain popularity.

Checkpoint notes:

I will be researching more specifics on the current iTunes video DRM system so I can present some more details on how it operates. I also intend to go into more detail about how if the iTunes video system ever actually intends to be highly successful, then it almost must rely on some form of P2P system in order to maintain the same download speeds during high demand as it does during low demand.

[1] First look: BitTorrent video download store

http://arstechnica.com/news.ars/post/20070227-8929.html

[2] Some observations on bitTorrent performance

http://portal.acm.org/citation.cfm?id=1064212.1064273

[3] BASS: BitTorrent Assisted Streaming System for Video-on-Demand

http://www.ece.ucdavis.edu/~cdana/MMSP05%20-%20BASS.pdf

[4] Analytical Model for BitTorrent-based Live Video Streaming

http://www.lk.cs.ucla.edu/PS/nime07f.pdf

[5] Anyone can broadcast video over the internet

http://portal.acm.org/citation.cfm?id=1167863

Class links

Incentives Build Robustness in BitTorrent - EconP2P '03 (the Bittorrent paper) http://www.bittorrent.org/bittorrentecon.pdf