Re: [DNSOP] Genart last call review of draft-ietf-dnsop-rfc7816bis-09

[Stephane Bortzmeyer <[e-mail]>]

On Sun, Jun 06, 2021 at 11:13:23PM -0700,
 Suhas Nandakumar via Datatracker <[e-mail]> wrote 
 a message of 72 lines which said:

> I am the assigned Gen-ART reviewer for this draft

Thanks for the review.

> Section 2.3
> 1. MAX_MINIMISE_COUNT and MINIMISE_ONE_LAB - are the values for these constants
> normatively defined or are they just recommendations ? Can the same be
> clarified in the document ?

The sentence "a good value is 10" seems to me indicating that it is
just a possible value. The important thing is to have a limit. Do we
think it should be rewritten with RFC 2119 words? MUST have a limit
and the RECOMMENDED value is 10?

> Section 4.
> The section starts with query for "foo.bar.baz.example" and walk through refers
> to a.b.example.org  as query input.  Also no reference to ns1.nic.example seems
> to be appear in the detailed flows.
>  Can this be updated it to match overall ?

Actually, there are *two* independant requests. One for
foo.bar.baz.example and one afterwards ("Here are more detailed
examples") for a.b.example.org. In the first one, ns1.nic.example is
indeed used.

Should we use the same QNAME for both?

> Section 5
> "QNAME minimisation may also improve lookup performance for TLD
>    operators.  For a TLD that is delegation-only, a two-label QNAME
>    query may be optimal for finding the delegation owner name, depending
>    on the way domain matching is implemented."
> This para doesn't clarify how the performance will be improved.  Can it
> be extended with some context around the same.

With QNAME minimisation, an authoritative name server MAY use exact
matching ("do I know foobar.example?") while without it, it MUST use
tree matching ("do I know thing.stuff.foobar.example or an ancestor of
it?") and tree matching is typically slower.