On 13/09/2024 08:41, libor.peltan wrote:
Dne 12. 09. 24 v 22:26 Petr Menšík napsal(a):
Also similarly, recursive resolver when talking to authoritative
server might use multi-qtypes single query instead of 3 separate
queries. It needs to cache that multi queries is supported. Similar
to EDNS0 support indication.
I'm quite an opponent of complicating the DNS protocol in order to
gain some optimizations. This all belongs to the '80s. Current DNS
servers can easily handle 3x the normal load of _legitimate_ traffic
like nothing. And if not, it's still cheaper to scale them than
inventing and implementing multi-qtype.
Libor
I disagree. We have a lot of battery operated devices nowadays. Sending
unnecessary queries separately requires additional energy wasted during
transmits and also during receives. I know, DNS packets are small and
there is a lot of much higher traffic wasting bandwith and CPU time. It
is not only about how many queries servers can handle. When it reaches
strong dedicated resolver, of course there it is just a little
optimization. But it multiplies per per request. Not every traffic
happens in data centers over multi-gigabit links. Some IOT clients may
save precious packets sent and received.
If the support is detected, it may make applications using stub a lot of
easier to debug. It needs just one response. No parallel queries
synchronization etc. If every client calling getaddrinfo() with
AF_UNSPEC generates 3 queries, it makes sense to me to join them into
single query (if supported). Especially unless it won't start anything
until all responses arrive (like getaddrinfo does). I know, browsers
will likely implement advanced logic. But other (simpler) applications
still exist.
Besides, it needs to be optional. You do not have to use it.
--
Petr Menšík
Software Engineer, RHEL
Red Hat, https://www.redhat.com/
PGP: DFCF908DB7C87E8E529925BC4931CA5B6C9FC5CB
_______________________________________________
DNSOP mailing list -- dnsop@ietf.org
To unsubscribe send an email to dnsop-le...@ietf.org
