MAC vs DUID issue

[sthaug at nethelp.no]

> > It isn't - it was given as an example of why, for some use cases, the MAC is a far superior identifier than some arbitrary identifier made-up by the client (and so not guaranteed to be invariable and unique).
> 
> Well, yes, except that there have been cases where many devices of the same type have shipped with the same MAC address.   So the mere fact that something has gone wrong in the past due to incorrect implementation of protocol requirements is not an argument for changing the protocol.


It is an argument for

1. Making the client MAC address available, and
2. Allowing IPv4/IPv6 address decisions to be made based on this MAC
address

IPv4 does 1 but not 2. IPv6, at least before the implmentation of RFC
6939, does neither.

Nobody has claimed that the MAC address is perfect. We *are* claiming
that in many cases (that we have experienced) the MAC address is a
*better, more stable* identifier than anything else. And that making
this information unavailable in DHCPv6 was a bad decision.

Steinar Haug, Nethelp consulting, sthaug at nethelp.no