reader config?

[Russ Allbery]

Brian Kantor <brian at UCSD.Edu> writes:

> I'm interested in using 2.3 when released as a leaf node system that
> will be supporting our entire campus news reading and posting needs -
> about 200 to 500 simultaneous readers, all humans (no 'pull' feeds).

> In general, our expiration pattern is fairly simple, where we expire
> binaries at one rate and most everything else at another, so the ability
> to do complex expiration isn't necessary.

> Is there a consensus on what storage configuration best suits this kind
> of use - i.e., should I use tradspool, CNFS with Berkeley-DB overview,
> what?

I'm getting ready to roll out the same thing here at Stanford, and I plan
to use CNFS for all binaries and tradspool for everything else, probably
with tradindexed overview for right now since the server can handle our
current incoming article load easily (and I have a separate transit
machine).  But Berkeley-DB is very tempting and I may change my mind.

-- 
Russ Allbery (rra at stanford.edu)             <http://www.eyrie.org/~eagle/>