From: tjacobs(_at_)redsword(_dot_)com
...
troubled by the notion that a stranger could authentiate itself merely
by preceding the mail message with a session key exchange.
Ah; that's because I suppose I wasn't as clear as I thought I was
about the purpose of what was proposed. It has nothing to do with
authentication, but instead is merely a mechanism to make the task of
sending unsolicited messages very resource (specifically CPU)
intensive, ...
That goal and mechanism needs no changes in protocols or MTAs. You
need only configure your MTA to refuse all mail except when the SMTP
client uses SMTP-TLS. That's easily done with modern sendmail. In
many and probably cases your MTA won't have a cert for the SMTP client,
and so the two MTAs will exchange session keys and then do the crypto
CPU grinding required for confidentiality. If you do have a valid
cert for the client, you'll also be able to authenticate the client.
This might work your CPU as hard as the spammer's, but that symmetry
or worse is unavoidable. The speeds of CPUs running MTAs on the
Internet today vary by more than a factor of 200, and probably more
than a factor of 10,000 if you include systems with hardware help for
encryption. It is impossible to burden the computers of spammers
without overtaxing some SMTP servers and some legitimate SMTP clients.
Vernon Schryver vjs(_at_)rhyolite(_dot_)com
_______________________________________________
Asrg mailing list
Asrg(_at_)ietf(_dot_)org
https://www1.ietf.org/mailman/listinfo/asrg