[IPv6]Re: Analysis of Ungleich ULA Registry

[Kyle Rose <[e-mail]>]

On Wed, May 22, 2024 at 9:21 PM David Farmer <farmer=
[e-mail]> wrote:

> Many of the names and organizations with the largest number of entries are
> generic entries. For example, there are 135 name entries with some version
> of "Admin" or "Administrator" and 52 organization entries with some version
> of "Private."
>
> Nevertheless, at least a handful of what seem to be real organizations in
> the registry purport to use more than 10 ULA prefixes. The limitations of
> the new known-local ULA list in draft-ietf-6man-rfc6724-update could be
> problematic for these organizations.
>
> Also, given the number of names and organizations with multiple entries,
> there seems to be some evidence for the need for ULA-C prefixes shorter
> than /48, especially if an organization's justification for shorter
> prefixes can be evaluated.
>

We are not the protocol police.

Organizations are de facto free to do whatever they want with ULA-R space,
including choosing prefixes shorter than /48 or choosing those prefixes in
a non-random way. The key distinction from many other protocol violations
is that the only possible harm is to themselves, should they in the future
need to merge with, or use ULA to route to, some other organization that
happens to use overlapping address space. I won't cry for them, but I might
chuckle.

I don't see how these use cases justify ULA-C or central assignment. I
might even suggest we go the other way and soften the language on ULA-R
prefix generation to make it clear that while they REALLY SHOULD NOT*
self-assign short prefixes or ones not generated at random, all operational
considerations resulting from making that choice are on them. It's purely a
foot-gun: you can't draw it from the holster or point it at anyone else.

Kyle

* Q.v. RFC 6919. This is a case for which one of those keywords might
actually be appropriate.