I'd like to know whether there are more modern statistics in answer to
FAQ0702:
Q0702: How well does Exim scale?
A0702: Although the author did not specifically set out to write a
high- performance MTA, Exim does seem to be fairly efficient. The
biggest server at the University of Cambridge (a large Sun box) goes
over 100,000 deliveries per day on busy days (it has over 20,000
users). There was a report of a mailing list exploder that sometimes
handles over 100,000 deliveries a day on a big Linux box, the record
being 177,000 deliveries (791MB in total). Up to 13,000 deliveries an
hour have been reported.
This is the first time that I've been the only person responsible for a
relatively high-load mail system. The system needs to be able to handle
(remote host failures aside) around 1500 messages a minute at times,
although the load will be much lower most of the time.
Obviously, these numbers are bigger than those given in the FAQ's
answer. However, I'm guessing (hoping) that the FAQ's answer is dated.
These servers are Dual PIII 1GHz / 1GB RAM beasts with fscking fast
SCSI disks in RAID packs, with Soft Updates on the /var partition. For
those of you who know FreeBSD, the servers complete a full 4.3-STABLE
buildworld in 30 minutes, 6 minutes longer than the known world record.
The reason I opted for Exim was that it has outstanding log files that
are easy (for me) to use to trace deliveries and failures, and also
easy to turn into useful management reports. The system I'm replacing
developed a culture of questioning the local mail server before checking
ANYTHING else. So the mail system needs to "prove its innocence" often.
So, as long as Exim's going to cope with the load, it's the right tool
for the job over here. I have most expensive network look-ups turned
off and split_spool_directory turned on. I'm confident that the box
will cope, but some external input would be reassuring. :-)
