Nigel Metheringham writes ("Re: [EXIM] Largest volume EXIM installation? "):
...
> Dan maintains that bandwidth conservation is pointless - bandwidth is
> cheap - and also questions whether there are really the gains that people
> claim by multiple rcpt to's. On mailing list systems the bandwidth saved
> is substantial - somewhere back in the exim archives there should be a
> message from me giving the figures for the exim lists. On our core mail
> systems I have to admit that this is all a bit marginal - bandwidth
> savings are probably of the order of a few percent (if you are really
> interested I could try to analyze). However the other advantages of
> knowing about all your addresses before you transport still kick in - we
> abuse systems that have had temporary outages less than a qmail system
> would (latency here plays off against network friendliness).
Even on a small multi-purpose system like mine the bandwidth savings
can be considerable. I host a mailing list with only 35 subscribers,
with 172 postings in the last 7 full days, totalling about 440Kb. The
mailing list is for the C.U. Rock Society, and nearly all of the
subscribers have University email addresses.
Qmail would have used 15Mb of bandwidth in the week to deliver the
mail to this list. Exim (counting `->' vs `=>' in the mainlog) used
about a _third_ of that. The whole system's outgoing SMTP traffic for
a week is around 20Mb, including 5Mb for this list, so using Qmail on
this system would increase the outgoing bandwidth usage of my mailer
by at least 50%. (That's not counting any of the other mailing lists
I host.)
Speaking as someone who pays for hosting bandwidth out of my own
personal pocket at European prices, I get very annoyed when North
Americans tell me that bandwidth is cheap and doesn't matter.
--
*** Exim information can be found at
http://www.exim.org/ ***
