At 01:24 AM 5/27/03 +0200, Markus Stumpf wrote:
On Sat, May 24, 2003 at 11:27:15PM -0700, Scott Nelson wrote:
Why would limiting number of recipients make a difference? With today's
computers and Internet connections cheap, additional SMTP sessions will be
initiated just increasing Internet traffic.
That's an interesting theory you've posited.
How, I wonder, could we prove it?
qmail only does one recipient delivery.
Statistics show (haven't an URL ready, sorry) that the overall "overhead"
against multi-recipient systems is nearly unmeasurable. Of course there ARE
situations, were it is a big gain to use multi-recipient strategy.
Most of the time recipient sorting on the sender site is rather CPU and
DNS expensive so in the same time some mailservers spend sorting
other actually finish the deliveries.
I'm guessing, but I think the qmail sorting problem is quite different
from the spammer sorting problem. Spammers don't have to sort on the
fly - they can sort their lists well before hand, and the list only
needs to be sorted once.
Still, if we knew the average number of recipients for spam
messages currently, and the average number for non-spam,
then we could at least make a reasonable estimate of the effects
of limiting it, in terms of number of connections, and total bandwidth.
If it turns out the average number of recipients
is less than 1.1, then clearly limiting it to one wouldn't
have much effect at all. If it's 50, then that's nearly a
50 to 1 increase in raw network cost. To fully understand
the ultimate effect we'd still need to know what effect raising
the cost of sending has on spam and non-spam, but "the average
number of recipients" is at least one of the things we "need to know".
It occurs to me that if the SMTP protocol was changed so that it
didn't /guarantee/ 100 recipients, it could still allow more than
one for trusted senders. (local policy would dictate of course,
but the I'd make the wording "SHOULD limit to one, unless the
sender is trusted") Established mailing lists wouldn't be affected
as much, since they would probably be whitelisted relatively
quickly by the big ISPs.
Scott Nelson <scott(_at_)spamwolf(_dot_)com>
_______________________________________________
Asrg mailing list
Asrg(_at_)ietf(_dot_)org
https://www1.ietf.org/mailman/listinfo/asrg