downside to 'minimal-responses yes;'?
Kevin Darcy
kcd at daimlerchrysler.com
Tue Feb 15 02:09:42 UTC 2005
Dave Sparro wrote:
>I'm running into some slow performance on one interface of my
>multi-homed caching server. Turning on minimal-responses seems to
>help out for now. What problems could I be creating by changing the
>default setting?
>
I imagine that non-minimal responses are mostly useful in situations
where you have a central caching server and a bunch of subsidiary
caching servers forwarding to it using "forward first" (or the
equivalent), as an attempt to optimize query performance. Since the
subsidiary boxes build up a lot of referral information from the
non-minimal responses they get from the central caching server, they can
recover morely quickly if the central caching server goes down or
becomes unavailable for whatever reason. With minimal-responses, those
same subsidiary caching servers will have to start "from scratch" if the
central server goes down, since they possess very little referral
information with which to work.
If you don't have a setup like that, I'm hard pressed to think of a
downside to minimal-responses. Certainly apps which are calling
gethostbyname() and/or gethostbyaddr() through regular stub resolvers
don't really care what's in the Authority or Additional sections of
responses...
- Kevin
More information about the bind-users
mailing list