Dave Crocker writes:
> On 7/21/2025 3:26 AM, Barry Leiba wrote:
> > The number of message recipients is
> > unrelated to the number of RCPT commands for the purpose of this
> > discussion, and the fact that messages often have many recipients does
> > not mean that MSAs and/or MTAs need to use multiple RCPT commands in
> > one SMTP transaction.

The number of recipients and number of RCPT commands are DEFINATELY
related. The number of RCPT commands will normally be either same or
smaller than the number of recipients unless one the addresses is
mailing list.

I.e., to be able to have multiple RCPT commands you do also need to
have multiple recipients, and those recipients need to be in the same
SMTP server or at least share the same next hop SMTP server.

As I said in the enterprise uses it seems to be really common to CC
lots of people inside the company to almost all email threads, and
depending on those environments there might be a lots of email
recipients going to the same next smtp host allowing the MTA to issue
multiple RCPT commands instead of sending emails as separate SMTP
transactions.

Quite often in those environments the initial hop going out from the
company is some kind of middle box that does all kind of things to the
email, for example adding warnings that this email is confidential, or
virus checking etc, and it would be beneficial for it to be able to do
those operations for every single recipient at the same time, i.e., IT
adminstrators in those environments will most likely try to make so
that emails are sent using one SMTP transaction with multiple SMTP
recipients.

> The earlier, extended discussion about this had various postings about 
> average number of recipients listed in the RCPT-To command of an SMTP 
> session.  The average number was barely larger than 1.0.

There are systems that will always split emails to separate smtp
sessions, and that will lower the average, but as I said in my
previous email I did see about 2-3% of emails going through iki.fi
having more one smtp recipient in one smtp session. 

> That is, the /observabl/e number of recipient addresses in nearly all 
> SMTP sessions was essentially 1.

Nearly all = 97% of time in internet in general. I would assume it is
much bigger if you take same statistics inside the enterprice system
that does not explictly split each emails to separate transactions.
-- 
[email protected]

_______________________________________________
Ietf-dkim mailing list -- [email protected]
To unsubscribe send an email to [email protected]

Reply via email to