An open-source version of the Napster server, ported to many operating systems (Linux, BSDI, Solaris, FreeBSD, IRIX, OS/2 and, of course, Windows). The project started in early 2000. The Napster protocol was reverse-engineered (protest from Napster followed shortly), and a simple server was implemented. The original Napster client does not allow users to pick the server they connect to. A third-party-tool, Napigator (www.napigator.com), resolves this problem, and most Napster clones don't have it.

Contrary to the Napster servers, OpenNap servers can index any type of files, not only MP3 music files. The basic principle with a single server is very simple:

  1. User logs into server
  2. User transfers list of own shared files
  3. User can search for files in remote index and gets results
  4. User can then initiate direct connection to other user and download desired file

Several OpenNap servers can be linked to each other. This creates a sharing environment that is less prone to failure. The linking must be confirmed on both ends, so a Gnutella-like mega-network cannot grow against the will of the admins. If a search is conducted on one of the servers in a linked "ring", it is transferred to all others. If the other servers have matches, they return them directly to the searching client. This is very bandwidth-efficient , but has the consequence that results can arrive in small "chunks" from each server in the network.

Long after Napster is gone or has turned into a large, evil corporation, OpenNap servers will continue to exist. They take little bandwidth (mind you, no MP3 file is ever uploaded to a Napster server, only the index), and even a user with a slow dial-up connection can run one, 24 hours if he wants to (find a DNS-mapping service at <http://www.technopagan.org/dynamic/>). Shut down this, RIAA!

The OpenNap homepage is at <http://opennap.sourceforge.net>.