Michael C Burns <[EMAIL PROTECTED]> writes:
> A fair number. Our cell (psu.edu) has just over 260,000 principals
> right now. It's running on an older IBM RS/6000 model 370, so it
> doesn't need alot of horse power or RAM (128MB). We had to upgrade from
> a Sun Sparc 1+ with 64MB RAM about two years ago as that started having
> paging problems. The db files on the DB server are about 50MB each now
> (kaserver, prdb). We don't have alot of disk space.
We've got something in the range of 50,000 to 60,000 principals last time
I checked, and our three authentication and VLDB servers are SPARC 10s at
present (soon to be upgraded to Ultra 1s just because we're retiring our
old SPARC hardware, not because they're too slow).
> We have about 35,000 volumes using 150GB on two fileservers. I've heard
> of other places having 10s of fileservers and over a terabyte of disk
> space.
We have around 20 active file servers and nearly 2TB of raw disk space,
although that's a bit high since we're still in the middle of migrating
things around between old and new storage. Currently around 700GB used in
87,000 volumes.
> We're moving this cell from AFS to DCE/DFS
I'm sorry.
> About four years ago when I first started managing this AFS cell we
> tried using three database servers, but to send a new copy of the
> databases from the master to the slaves took 40 minutes or so. FTP was
> faster than whatever mechanism AFS was using to do the updates.
Never seen this problem. We've been running with three VLDB servers for
years (although we're currently temporarily running with five, soon to be
seven, while we do an IP address renumbering).
--
Russ Allbery ([EMAIL PROTECTED]) <URL:http://www.eyrie.org/~eagle/>
