Re: IETF SMTP Working Group Proposal at smtpng.org

[Brad Knowles <brad.knowles () skynet be>]

At 7:43 PM -0400 2002/08/22, Barry Shein wrote:

>  If you want slower e-mail delivery why not just put sleeps into the
>  receiving MTA?

	That's basically what postfix does.  Based on the types of abuse 
(e.g., too many bounces in a given period of time), it goes into a 
bounded exponential backoff for the amount of time that it will sleep 
between operations.  I am proud to say that I was a champion of this 
kind of behaviour, and I helped convince Wietse that this was a good 
thing to do.


	However, what we're talking about here is a way to know, a 
priori, whether or not you should start immediately sleeping long 
periods of time between transactions, based on a "cookie" that is 
transmitted early in the SMTP dialog.

	If you've got the cookie, then there is no extra sleeping added. 
If you don't, then there are added sleep loops for the very first 
message and all following messages (maybe you would even refuse to 
accept more than one message per connection, slowing them down even 
more).

	The only way to get the cookie is to be a good netizen and sign 
up with a well-maintained central white list operator, who would then 
issue you the cookie in question.


	What becomes interesting in this kind of case is handling the 
cookie distribution infrastructure, as well as the cookie revocation 
infrastructure.


	However, all of this is way off-topic for NANOG, according to the 
proposed FAQ.

--
Brad Knowles, <brad.knowles () skynet be>

"They that can give up essential liberty to obtain a little temporary
safety deserve neither liberty nor safety."
    -Benjamin Franklin, Historical Review of Pennsylvania.

GCS/IT d+(-) s:+(++)>: a C++(+++)$ UMBSHI++++$ P+>++ L+ !E W+++(--) N+ !w---
O- M++ V PS++(+++) PE- Y+(++) PGP>+++ t+(+++) 5++(+++) X++(+++) R+(+++)
tv+(+++) b+(++++) DI+(++++) D+(++) G+(++++) e++>++++ h--- r---(+++)* z(+++)