I would personally like an AS112 type registration. a well known anycast IP
range, which has letsencrypt HTTPS marking for the IP as numbers, so we go
to https://2001:db8::53/ as an endpoint and get the closest anycast service.

I like the majors, and I like them offering things like 1.1.1.1 and 8.8.8.8
but I think if we're exploring a future of the fetch of the root, I'd
rather an AS112 type thing was offered by the majors, amongst themselves,
than a single major said "we'll do this for community benefit"

(again, this is personal opinion and nothing else)

-G

On Thu, Jan 22, 2026 at 1:43 PM Wes Hardaker <[email protected]> wrote:
>
> George Michaelson <[email protected]> writes:
>
> [adding gas to the fire....]
>
> > I'm broadly comfortable with the document separation but it does beg
> > the question what is the impact of one or other of them being blocked,
> > or left to rot? Because if we're in an all-or-nothing world it needs
> > to be called out (I am pretty sure we aren't btw)
>
> Well, that's fair though to some extent some of them are independent
> (eg, the URL scheme should probably be done regardless) and some of the
> separation, as I said, is to separate the discussion a bit into the very
> different components.  We may want to merge them.  Or not.  But we could
> fail the two IANA ones and the main LocalRoot document and the URL
> scheme documents are still likely helpful on their own.
>
> > I am least comfortable with the last one. I think IANA would not want
to be
> > given the task of judging suitability to list.
>
> Well, that's actually part of the goal of providing guidance about what
> types of providers should go in there.  Some people I've talked to think
> we should just trust IANA to build the list.  I think "some" guidance
> would be good.  But it could be a more standard process too.  And I
> agree this is a larger component of the discussion.  I do need to have a
> (another actually) good heart to heart with Kim too.
>
> But the reality is that with ZONEMD and DNSSEC in play, it really
> doesn't matter how you get the contents as long as you can.  Though the
> document doesn't say or suggest you can use carrier pigeons, you
> certainly can (assuming a good fleet of pigeons that make regular
> trips).  The providers of the data need to be robust, but as long as you
> can reach some you should be good.
>
> Yes I'm hand waving a lot, but the underlying goal should be: the IANA
> root zone data provider list should be easier to fill and change than
> the existing root server addresses that actually serve the root zone.
> There aren't privacy trust issues. It can be longer.  They only need to
> be queried occasionally.  They can reuse existing well established CDN
> infrastructure.  And if one fails, you shouldn't care as much if others
> work. A complete failure to contact any due to local outages won't
> matter as much Just make sure you use and trust the object protection.
> IANA needs to house a default list.  But operators can build their own
> list if they like.  Just check the data.
>
> --
> Wes Hardaker
> Google
_______________________________________________
DNSOP mailing list -- [email protected]
To unsubscribe send an email to [email protected]

Reply via email to