[ih] Fwd: Design choices in SMTP

Dave Crocker dhc at dcrocker.net
Tue Feb 7 18:23:42 PST 2023


On 2/7/2023 6:06 PM, John Day via Internet-history wrote:
> As John points out, mail was originally part of FTP and mailboxes were files not directories. If one had multiple pieces of mail to deliver to the same mailbox would they have been done individually or all at once?

The FTP mechanism was one addressee per transfer.  As I recall, the 
major incentive for doing SMTP was to allow a transfer to be to multiple 
addressees.

The UA/MTA distinction, with separate software doing user interaction, 
from the software that did transfer, started appearing in the latter 
1970s.  By 1980, there were enough examples of that architectural 
approach to justify a pre-X.400 effort (IFIP WG 6.5) that introduced the 
UA/MTA construct.

Given that the Arpanet and corporate networks like Xerox's and Digital 
Equipment's had fulltime connectivity among hosts, the operational 
approach was to send immediately upon user submission, doing queuing and 
retrying only after initial failure. Given the paltry traffic load of 
those days, it made no sense to wait and send a batch to the same place, 
since there was no benefit in waiting and no cost differential in 
packaging mail to a site together.

The exception was telephone-based transfers, given the cost of 
long-distance calls in those days.  And indeed, phone-base mail relaying 
systems DID batch messages to the same site, to get time-of-day 
benefits.  Some also allowed mixed connection initiation, so that either 
side might connect when rates were best for them.  Then any waiting mail 
would flow as needed.


>   If so, when did delivering mail one at a time start? In SMTP? When mailboxes became directories?

see above.

It is a bit ironic that bulk mail these days typically is sent one 
addressee per transaction, due to content customization.


> (I would also note that Multics was uneasy giving an anonymous user Write access to a file so created Append access for use by mail.)

Giving any daemon privileges is always a risk. As I recall, sendmail 
gave superuser status to the entire program.  For MMDF I only gave it to 
the local delivery process.

d/

-- 
Dave Crocker
Brandenburg InternetWorking
bbiw.net
mast:@dcrocker at mastodon.social




More information about the Internet-history mailing list