[arin-ppml] IPv6 policy: limitations on registrants?

[John Osmon]

On Sun, Jun 28, 2026 at 06:04:42PM -0700, William Herrin wrote:
> On Sun, Jun 28, 2026 at 2:04 PM John Osmon <josmon at rigozsaurus.com> wrote:
> > On Thu, Jun 25, 2026 at 03:29:15PM -0700, William Herrin wrote:
> > > Should they have to be for use on the Internet with off-Internet uses
> > > expected to use something like ULA? Or are off-Internet uses just as
> > > qualified as Internet uses?
> >
> > I'm a proponent for unique address space at all times.  Even an
> > "off-Internet" network might need to interconnect with a new network
> > (or, even, the Internet).
> >
> > Allowing unique address assignment means that if (when) such
> > interconnection occurs, the formerly private network can coexist.
> 
> Hi John,
> 
> I take your statement to mean you favor off-Internet use being a valid
> justification for ARIN IPv6 address registrations. Is that correct?

sorry for the ambiguity:

I'm very much in favor of the issuance of registered/unique address
space for people that wish to use IP, even if they aren't planning on 
connecting to other networks.

 
> Bear in mind that IPv6 Unique Local Addresses under RFC 4193 are not
> the same as RFC 1918. Like the rest of IPv6, the address space is much
> larger and as the name implies the intent is that few if any users
> select the same addresses. A simplification of the ULA rules boils
> down to: generate a random number between zero and a trillion and then
> add "fd" at the front. That's your /48 within fd00::/8. The birthday
> paradox means overlap is not as unlikely as it might seem, but overlap
> is still pretty unlikely.

Yep.  I've seen:   fd00::/48, followed by fd01::/48 rolled out because
"it's easier to remember and type."

I often wonder how many fd00:dead:beef::/48 networks exist...

 
> That said, having ARIN assign an address block guarantees that you
> won't accidentally overlap without relying on any stochastic foibles.

Stochastic human action seems to be predictable when you don't want it to
be...




> With this additional information, what are your thoughts?

I favor "off-Internet" use as justification for address space, from ARIN
or any other RIR.

I reserve the right to change my mind as 2000::/3 dwindles, but until
then I think I will remain in this camp.