Re: [Exim] Re: Very large mailing lists (Exim vs Isocor)

On Mon, 10 Apr 2000, Dave Cinege wrote:

> > 30K messages/hour is only 8 messages/sec. Our lists (using qmail) average
> > that when they are going, and peak out at 13msg/sec. This is all on a P133
> > w/64M ram and IDE disks. Primarily we are disk bound (on the fsyncs of the
> > user list)
>
> Damn....what OS, and does it do anythign else besidesact as MTA?

Linux (Debian actually). It does deliveries and smartlist - smartlist
actually consumes more resources than qmail does, that is why our average
is only ~8/sec and peak is upwards of 13.

> Exim can be set to spawn up to X amount of processes per queue and up to
> a max of X concurrent queues during delivery. Does this not put it on
> par?

There is no reason Exim couldn't reach the peak delivery rates, but on the
latency and boundedness (ie number of exim processes) I think it has to
loose out to the mailers with central dispatch schemes. I've never
actually had qmail run away on me, it neatly goes up to the limit I set
and sits there until it is done.

Jason