Build a P2P Black Box

Peer-to-peer (P2P) technologies have received a black eye for their popular yet illicit uses. But as the P2P technology industry has been saying all along, there are many legitimate applications for file sharing, too. Large archives can be efficiently distributed to a widespread audience with P2P, and when downloaders participate in the upload process, they help to reduce strain on central servers and increase the overall bandwidth available to those who wish to acquire the files.

Arguably, the best technology today for distributing a focused archive of content to a widespread audience is BitTorrent, often simply called "torrent." Unlike many other P2P technologies, torrents do not rely on unstable or seedy distribution servers. Using BitTorrent, a company can run its own clean, members-only distribution network, cooperatively spreading data with great speed.

In this TechBuilder Recipe, I'll show you how to construct a dedicated Linux server to act as a sort of P2P black box. The server will act as a self-contained torrent-serving machine.

Torrent Lingo

You'll understand this process better--and be better able to explain it to clients--if you have a basic command of BitTorrent vocabulary. To get you started, here are the basics:

• Peer: Also known as a "downloader" or "leecher," a peer is a participant in a torrent network that is currently downloading a file, and that has less than 100% of the complete archive. Peers are expected to allocate some of their bandwidth to uploading the fractions of the archive they do have to other peers. BitTorrent clients manage this relationship automatically.

• Seed: A seed is a torrent peer that has 100% of the torrent archive and is uploading it to others. The P2P black box will act as a torrent seed, but other peers may become seeds if they complete the archive and choose to remain connected to the torrent.

• Tracker: A torrent tracker is a small piece of server code that acts as an information clearinghouse for the seeds and peers. The tracker tells the seeds and peers both the parameters of the file—how large it is and how many chunks it has been divided into--and how seeds and peers can locate each other and attempt to connect. The tracker will run on the P2P black box as part of our BitTorrent client.

For additional reading on torrent technology, visit the official BitTorrent FAQ.

Ingredients

Here's what you'll need to build this system:

• A 21st-century server: Although torrents are relatively efficient, our recipe will use a Java application, and our torrent may be managing hundreds--or even thousands--of users. So you will need a machine with at least 1 GHz in CPU power and 512 MB of RAM. Any 100BaseT or better Ethernet connection will suffice. Also required is a 7200-RPM hard drive with at least 5 GB more than the volume of data that will be served. A DVD-ROM drive is recommended for ease of OS installation, as well as for loading data to be served via torrent.

• Server OS: Linux and its cousins continue to be the most cost-effective choice for this type of dedicated-purpose, fire-and-forget box. For this recipe, we will use Fedora Core 4, an offshoot of Red Hat Linux, for its flexibility and comparative ease-of-use with Java.

• Java Runtime Environment: Available from Azureus as the Torrent server application. Azureus is graphical and easy to use, and it contains its own tracker and mini-Web server, making the server easy for customers to manage in the field.

• Data to be served: Although a torrent can be used to distribute files of any size, you'll generally find it most effective and efficient for substantial archives, in the 10 MB-and-up range. Applications or large application patches, video captures, database dumps, and the like all make prime candidates.