Re: [v6ops] Implementation Status of PREF64

[Lorenzo Colitti <[e-mail]>]

On Fri, Oct 1, 2021 at 9:35 AM Owen DeLong <owen=[e-mail]>
wrote:

> That’s not what is being said. A lot of this has little to do with what
> they are accustomed to and a lot more to do with a combination of:
> + Enterprise policies that they cannot conveniently change
> + Government and other regulations that they cannot avoid
> + Reporting requirements that they cannot subvert
> + Limitations of the tools they have available to comply with the above.
> etc.
>

Ok, but many of these points can be addressed in other ways:

   - For reporting: many vendor platforms support ND binding logging via
   syslog, and something like draft-ietf-dhc-addr-registration
   <https://datatracker.ietf.org/doc/draft-ietf-dhc-addr-registration/>
   would likely make it much easier for networks that use DHCPv6 only for
   tracking/forensics (vs. for access control) to move away from IA_NA.
   Specifically, it would satisfy the reporting requirements and make it a lot
   easier for the tools to adapt because it's still using the DHCPv6 protocol.
   - Enterprise policies and government regulations are not written in a
   vacuum, they're written to address business and legal requirements, and I
   really don't think you can credibly claim that DHCPv6 is a business or
   legal requirement. The legal requirements are things like tracking, access
   control, etc.



> In other words, you keep trying to suggest that there are alternative
> technical solutions, but you ignore the fact that this is NOT a technical
> problem.
>

Actually what I'm saying is that we (the IETF) *do* have a technical
problem, regardless of the status of implementations, which the IETF
doesn't have much of a say in anyway. The problem is: there is no IETF
consensus that "IA_NA with one address per device" is a good technical
solution. So: why aren't we looking for a way that's better for everyone
involved (users, implementers, operators)? Even if you're 100% convinced
that utimately all platforms will implement IA_NA with one address per
device, isn't it a better outcome if that model is limited to a few
networks, and networks with less stringent requirements or less dogmatic
operators can use a technically better solution that we come up with in the
meantime?

Of course, there may still remain network operators that say, "my network,
my rules; my one IA_NA way or the highway". But if your main concern is
accelerating IPv6 adoption, then if we come up with a solution that's not
IA_NA and satifies some (even if not all) enterprise requirements, then
that's pretty likely to help adoption of IPv6 in enterprises that have a
real business need for IPv6.