[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: fooling MTAs.



oh come on - how about simueltanous mail - that would tend to cut in on
your accounting...
SMTP is the dominant method of int'l E-mail these days
the overhead it incurs on already used lines is very low
(even for say 16 sessions in parallel ) unless you're using E-mail for
file transffer
in which case it is now worrse ( rorder of magnitude ) than say - FTp
( keaving aside say MIME encapsulation )
in any case proper MTA will also transfer ,multiple mesages - each of
which may have multiple recipients

no how many of the 6 below can be ppre-loaded using TCP streaming ?
don't forget most MTA's only check for an error for the RCPT & at the end
of DATA ( as ppl who try to write MTA that abort due too to large a
message or invalid recipients know too well )
I can't bother to check your logic for any other flaws since
obviulsy SMTOP although not optimal does wok long haul ...
---------------------------------
HELO
MAIL From:
RCPT To:
DATA
.
QUIT
-------------

-- 
Rafi Sadowsky                                   rafi@oumail.openu.ac.il
Network/System/Security  VoiceMail: +972-3-646-0592   FAX: +972-3-646-5410
       Mangler ( :-)      |            ILAN-CERT(CERT-L@VM.TAU.AC.IL)
Open University of Israel |     (PGP key availble by finger or key servers) 


On Wed, 27 Aug 1997, Peter wrote:

> 
> >body and envelope
> 
> Ok, I know that. And I can almost understand about the thing about the 
> SMTP server friendlily warning about 'may be spoofed'. What I don't 
> understand is, why does it accept the message if the From: domain is not 
> the same as the IP of the incoming connection resolved by the name server 
> ?! Normally I would enforce dropping connections with the From: domain not 
> resolving to the sender IP (or an accepted sender from an ACL).
> 
> Out of line: I was thinking about SMTP and long haul mail transmissions. 
> You know, there is a thing called ping time, the time a radio signal
> requires to fly up and down to/from a sattelite or pass through zillions
> of wavelengths of fiber or copper wire. In SMTP the smallest number of
> question-reply pairs required to pass a connection is of 6. This means
> that the smallest mail message (null body) passes through a channel of ANY
> speed (no matter how fast) within MORE than 6*ping_time. 
> 
> If the ping_time is 0.3 sec (optimistical value over the Atlantic Ocean,
> i.e. from Israel to the States) then the shortest time required to pass 
> an email via SMTP is of 1.8 seconds. With 10,000 messages to pass this is 
> 18,000 seconds, 300 minutes or 5 hours. 
> 
> I infer that *NOONE* (sane) is using SMTP over long-haul. So what is being
> used ? A variation of UUCP batch mail over TCP ? BTW, noone could afford
> to send email if he had to pay for 1.8 seconds of sattelite channel air
> time for each message. 
> 
> subject for rumination....
> 
> regards,
> 	Peter
> 	(plp@actcom.co.il)
> 
>