Re: [Asrg] 6. Proposals - Let pull mail replace push mail

[Brad Knowles <brad.knowles@skynet.be>]

At 11:08 AM +0200 2003/10/01, Dag Kihlman wrote:

> >  Please search the archives.  This has been discussed before.
	Indeed.

> >  1. It greatly increases the traffic on the internet.
>  If you mean more connections that is true, but if you make a calculation
>  you will find that the increased traffic is nill and nothing compared with
>  the
>  traffic caused by spammers, viruses not to mention radio listners and
>  video viewers. Compared with this traffic the traffic argument is not really
>  a big thing. Historically the argument is valid but it is getting less and
>  less
>  interesting.
	Additional traffic is not the only problem.  What was a single 
e-mail message is now a message body that has to be stored somewhere 
reliable, plus a notification to the recipients that they can come 
pick up their e-mail at such-and-such location.  What was a single 
e-mail message now has two parts, which increases by an order of 
magnitude (or more) the problems with things breaking down and 
getting disconnected or mis-directed.

	Now all spammers would have to do is to compromise a mail silo 
server somewhere, replace all the message bodies with their spam 
content, and they can be assured that this will get past all the 
recipient filters because the original notices that went out were 
real e-mail messages from real people.

	In addition, if you've ever built a large-scale mail system, you 
know that the single biggest problem facing you is not disk storage 
capacity, but is disk I/O latency, especially with small messages. 
You're talking about taking all traffic and turning it into a message 
upload plus one or more small messages.  This is the worst possible 
thing you could do to the mail system.


	If you want to invent the Mail-Merge Transport Protocol, talk to 
Eric Allman and Bryan Costales.  We went a long ways towards 
designing this thing while I was at AOL and Bryan was at 
MercuryMail/InfoBeat.

	Problem was, we realized that this would increase the spam 
problem by many orders of magnitude -- spammers already use tools 
that send the message body once and then a list of recipients to 
which to spam that message to, we'd just be allowing them to push 
that mechanism to the end recipient.


	See the archives.  Please don't make blind proposals like this 
without having reviewed what proposals have already been made, and 
have some suggestions for how you're going to address the problems 
previously identified.

--
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(+++)

_______________________________________________
Asrg mailing list
Asrg@ietf.org
https://www1.ietf.org/mailman/listinfo/asrg