> Tony Finch <mailto:d...@dotat.at>
> Tuesday, January 06, 2015 2:16 AM
>
> ...
>
> So if you use an NS query to probe zone cuts, a straightforward
> implementation of query minimization requires four queries for a three
> label name (if the resolver has a primed but empty cache).

that's true in an empty cache. but that emptiness will be short lived.
at stead state, query-minimization has no more queries than longest-match.
>
> If you use A queries you save a lookup in the common case where there
> isn't a zone cut at the leaf.

since the intermediate queries (all but for the qname itself) are
intended to solicit referrals, the qtype on those doesn't matter. they
should probably use the original qtype, because of middleboxes who would
see qtype=NS as unnatural.

-- 
Paul Vixie
_______________________________________________
DNSOP mailing list
DNSOP@ietf.org
https://www.ietf.org/mailman/listinfo/dnsop

Reply via email to