John C Klensin wrote:
--On Friday, 06 July, 2007 11:53 -0700 Douglas Otis
<dotis@xxxxxxxxxxxxxx> wrote:
This complex topology spells the end of SMTP in its current
form.
...
Doug, I think you are conflating two problems. There is running
code (and extensive history) to demonstrate your conclusion is
not correct;
As you note, the world used to be rather more complicated and it worked fine.
Besides the integrate service, using heterogeneous email standards, that you
cite, folks might want to take at another historical reference:
<http://www.isi.edu/in-notes/rfc1775.txt>.
That the current world is pressing for end-to-end authentication is not a new
requirement. The challenge is present whether we have v4/v6 or just one IP,
to serve us all.
But, then, email is multi-hop -- contrary to some people's model -- so that's
why we try to design mechanisms that work... multi-hop.
On the other hand, any authentication, authorization, or
validation technique that depends either specifically on IPv4
addresses or on some sort of end-to-end connection between the
This is only one of the examples of the fundamental weakness of any scheme
based on Path Registration -- that is, authentication based on a topology tag,
such as an IP Address. Path Registration works for single-hop scenarios,
albeit even then with significant limitations.
Specifically: tying topology information to messaging means that it cannot
work for multi-hop scenarios.
d/
ps. We can have some fun debating the meaning of "multi-hop" since travel
through a sequence of IP routers is multi-hop, even though the TCP connection
is "direct". Similarly, an SMTP or SUBMISSION session that goes through a
proxy might entail multiple TCP connections, but the mail-level view is that,
again, the interaction is direct.
--
Dave Crocker
Brandenburg InternetWorking
bbiw.net
_______________________________________________
Ietf@xxxxxxxx
https://www1.ietf.org/mailman/listinfo/ietf