[DNSOP] Re: Embedding addresses - draft-wobbly-dispatch-embedded-answers

[Ben Schwartz <[e-mail]>]

I am totally supportive of this but it is a much bigger project than this
document suggests.  Cryptographic validity is not enough; we need proof
that the record is intended for this client at this time for DNS load
balancing.  Imagine that a popular search engine messes up its
"resolverless" configuration and starts sending everyone who searches for
your website to a single random node in your CDN...

There are various ways we could try to solve this, but they generally
require new DNS protocol elements _and_ new web browser behaviors, so the
barrier to deployment is very high.  Also, we need to benchmark against
<link rel="dns-prefetch"> and all the other latency-hiding tools we already
have.

--Ben

On Fri, Oct 30, 2026, 7:17 PM Warren Kumari <[e-mail]> wrote:

> Dear DISPATCH, DNSOP & HTTPBIS[0], Joe and I would really appreciate your
> review and feedback on this draft:  draft-wobbly-dispatch-embedded-answers
> - "DNS Embedded Answers" This idea is strong influenced by the
> "Resolverless DNS"
> 
> Dear DISPATCH, DNSOP & HTTPBIS[0],
>
> Joe and I would really appreciate your review and feedback on this draft: draft-wobbly-dispatch-embedded-answers
> - "DNS Embedded Answers"
> <https://urldefense.com/v3/__https://datatracker.ietf.org/doc/draft-wobbly-dispatch-embedded-answers/__;!!Bt8RZUm9aw!_EzycVzKC97e3Z7aHg0ByqrrFi6ZBrvDtrJrlXwqJWEoeYs3RipBGsN5UVsu1CcAw0_OzSir0gA$>
>
> This idea is strong influenced by the "Resolverless DNS" discussions from
> 2018.
>
> We believe that the general concept deserves more discussion, and so we
> have written it up as a draft.
> Please keep an open mind - the emotions ran somewhat strong during the
> initial Resoverless DNS discussions. This is written up in the original
> spirit of request for comment - we'd like feedback and discussion on the
> idea and concept, as well as on the document itself. It clearly needs more
> work, but we'd like to either understand how to make this work, or why it
> won't.
>
> I have done some initial performance testing - depending on the network
> environment, resolver, phase of the moon, etc it seems that on average DNS
> is taking up between 0.5 - 6% of my browsing time (time from initial
> connection until Largest Contentful Paint (LCP)).
>
> We believe that it would be great for the user experience if we can reduce
> that, providing it can be done in a privacy (and security) protecting
> manner.
>
> Please help us understand if we can do this, or if this is a bad idea,
> Warren and Joe.
>
> [0]: Sent to all three list desperately to cut down on cross-posting.
>
> _______________________________________________
> DNSOP mailing list -- [e-mail]
> To unsubscribe send an email to [e-mail]
>