Thanks!

I used the Hunchentoot web server, running on SBCL on an Ubuntu server (1GHz Athlon, 1GB ram, residential cable connection).

The data-sources are html files, which are sucked into memory when lispdoc starts up. The various indexes are kept as hash-tables in memory - there's no need for a database since everything fits in ram.

What is your experience with Hunchentoot so far? It sounds pretty impressive from the specs.

Hunchentoot is awesome - very comprehensive and very easy to work with. Web development has never been so easy (before I switched I was using Araneida for most things, and mod_lisp for a few others).

I can't say I've tested it with huge loads, but it's been very stable so far.