On Fri, 9 Jul 2010, Tony Li wrote:
In the real world, MAC addresses are not unique, they're just unique
most of the time. Any scheme that depends on unique MAC addresses need
to gracefully handle when they're not.
In the real world, there are no perfect solutions. Why is everyone
insistent on an architecture that solves all problems, even in the face
of equipment that clearly does NOT follow existing architectures?
When I designed a ADSL residential broadband solution in 2002 (or so) I
made sure we could handle duplicate MAC addresses between customers (this
was solved by RFC3069 capable equipment) and I was very happy about that
when we discovered a few years later that 5% of our user base was using
the same MAC address.
I don't know enough about the pros and cons of each proposed solution
(because I haven't really found an easily understandable
summary/presentation), but I'm saying that operationally (working at an
ISP) I want any solution proposed to gracefully handle identifier
collissions (either intentional or unintentional).
It took a long time before there were standards/proprietary mechanisms
that made it possible to deploy IPv4 securely to residential customers
(finally we now have the SAVI-WG that'll hopefully make this happen for
both IPv4 and IPv6) and any solution proposed for the future should have
this kind of work built in already.
--
Mikael Abrahamsson email: [email protected]
_______________________________________________
rrg mailing list
[email protected]
http://www.irtf.org/mailman/listinfo/rrg
