conor@??? said:
} 1. How scalable is EXIM, i.e are there any other ISP's/Organisations
} using EXIM to serve upwards of 40,000 customers ? If so how does EXIM
} compare to sendmail in this environment.
How long is a piece of string....?
There are a number of organisations with this sort of customer base - but
the customer base is not one of the vital figures related to sizing the
system, but rather throughput - messages and volume per hour/day.
Quoting from an earlier message of mine:-
> Issues include how your mail is distributed (a single mail spool, maybe on
> one box, maybe on networked file system, or split across machines and how
> you load balance), access to the mail (direct file, pop, imap, web,
> other), mail spool formats, hardware types etc....
>
> I think that exim is relatively light on CPU and system resources. I
> don't know how it compares to other systems, but my gut feeling is that it
> is at least comparable to most and better than many. However what works
> for me may not in your situation.
>
> We use a system based on multiple relatively low powered servers front
> ending an NFS mounted mail spool (on a *big* NetApp which is hideously
> underused). Scalability of that system depends entirely on the backend
> file system - we can just add more front end boxes ad-infinitum. I did
> design most of the system to be very network file operation efficient...
> this may not apply so well to other systems. With the volumes you are
> talking about, I would have no problems at all (allowing for adding a f/e
> box or so). The f/e boxes are SPARC20's - ie about the power of an intel
> P75 :-(
>
> If you are looking at a single big box approach then you will have very
> different requirements and trade offs to me.
Using a pipe transport will also cost, although not as much as you might think since exim tends to fork/exec even for a local file delivery. Its still an overhead.
} 2. I have been attempting to get EXIM to use a binary called spop to
} perform local deliveries. I have been attempting this using a pipe
} transport that calls this binary and naming that transport in the
} aliases director.
} command = "/lib/mh/spop ${lookup {${local_part}} lsearch {/etc/aliases} {$value}}"
make that
} command = "/lib/mh/spop ${lookup{${local_part}}lsearch{/etc/aliases}{$value}}"
The parser should be smarter, but isn't...
However I wonder why you have carried the alias lookup to the transport.
Why not redo the user in the director, and the just use $local_part in the
transport - what would happen here if the lookup fails in the transport??
Also I find it strange that you talk about a large volume installation and
then use lsearch lookups, unless this is a very small list of users...
Nigel.
--
[ Nigel.Metheringham@??? - Systems Software Engineer ]
[ Tel : +44 113 207 6112 Fax : +44 113 234 6065 ]
[ Real life is but a pale imitation of a Dilbert strip ]
--
*** Exim information can be found at
http://www.exim.org/ ***
