The proposals here (BLOB, CONTAINER, BINARY, etc.) are significant
changes to the DNS system design, comparable in scope to the TXT
record.  I do not think that major changes like this should go through
IANA "Expert Review" without IETF review.

In general, the DNS has historically favored somewhat purpose-specific
RR types.  In my view, the design goal is that general-purpose DNS
tooling should understand the data well-enough to render it in
human-readable fashion, to the extent possible.

A BINARY type would lose this ability, and encourage arbitrary opaque
data in DNS.  Perhaps this is what we want, but it is a big change
with some obvious downsides related to debugging and inspection.

Personally, I would be more comfortable with a "CBOR" RR type or
similar, as this at least offers the possibility of human-readable
visualization and editing.

--Ben

On Wed, Oct 28, 2026 at 11:44 AM Mukund Sivaraman <[email protected]> wrote:
>
> >
> Hi Ray
>
> On Wed, Oct 28, 2026 at 11:37:06AM +0000, Ray Bellis wrote:
> > [*] it's an omission from the DNS RRtypes that I've noticed before,
> > with the closest being NULL, but this type is not permitted within
> > zone files.
>
> Even I was wondering about NULL. It seems NULL was used for completion
> queries which was deleted from RFC 1035:
>
> >  The optional completion services described in RFC-882 and RFC-883
> >  have been deleted.  Redesigned services may become available in the
> >  future.
>
> It was used for name completion and had don't-care RDATA in such queries
> (it was not from zone data). There is an example of using it in name
> completion in RFC 833 on page 22.
>
> It is strange that a BLOB-like RR type has not been registered so far.
>
> @Guorong Tian: You might also consider using an RR type in private-use
> space in RFC 3597 presentation format with a owner name prefix. It would
> be essentially the same thing as using a BLOB RR type to store arbitrary
> data. It would be problematic if a NS implements some private
> experimental functionality in that space and expects it to have some
> specific RDATA structure in its wire parser, but proper DNS
> implementations parse RDATA of private-use space as opaque and leave
> structural parsing to where it's used.
>
> You may also want to look at Eastlake's SINK RR type which was meant to
> be a kitchen sink of container types (because you wanted to register a
> CBOR container type):
>
> https://urldefense.com/v3/__https://datatracker.ietf.org/doc/html/draft-eastlake-kitchen-sink-02/__;!!Bt8RZUm9aw!8ivJxPe7QpxDToVg9zZtaw2EKX62WyrPfvKJ2PXQGLJ4XoKXmaB-bhlxilOpl3vKN3qyNDK-$
>
>                 Mukund
> _______________________________________________
> DNSOP mailing list -- [email protected]
> To unsubscribe send an email to [email protected]
_______________________________________________
DNSOP mailing list -- [email protected]
To unsubscribe send an email to [email protected]

Reply via email to