Date: Wed, 18 Feb 2004 11:35:54 -0800
From: Dave Crocker <dhc(_at_)dcrocker(_dot_)net>
I should comment that I think that the slide has far too much detail,
for most discussions. Most of the time, folks are not bothering to
look at _any_ of the tradeoffs, so a diagram that detailed can get
distracting.
On the other hand, I think it is useful for helping
people to see just how complicated email can be. With luck, this will
reduce the tendency to treat it simplistically.
Also on the other hand, the chart omits some detail that is critical to
some discussions. For example, there have been some recent discussions
elsewhere regarding DSNs and bounces provoked by some of the recent
worms (http://www.imc.org/ietf-smtp/mail-archive/msg01062.html
http://www.dshield.org/pipermail/list/2004-February/subject.php (Subject
"[Dshield] SPF is fundamentally flawed" (it's not primarily about SPF)
(the links were broken when I tried them this morning)).
As I understand it, the "Bagle" worm used an internal SMTP engine to
send directly to the recipient's MX host (flow 1 in Patrik's diagram).
In the case of a non-existent recipient in that domain, the MTA (presumably
MTA(r1) in Patrik's diagram) would send a 5yz response and there would
be no bounce or DSN. However, it appears that a number of sites use a
"perimeter" MTA which does not check for recipient validity, but instead
stores and forwards to another MTA in the recipient's domain which does
perform such a check, and which is then responsible for a DSN or bounce
(which goes to the forged sender envelope address).
_______________________________________________
Asrg mailing list
Asrg(_at_)ietf(_dot_)org
https://www1.ietf.org/mailman/listinfo/asrg