Performance hit on Query logging

[Eivind Olsen]

--On 7. oktober 2010 16.55.54 -0500 groups <groups at obsd.us> wrote:
> One party thinks that disabling query logging will give enormous
> performance gains, while 30% is a lot.. IMHO it is very negligible in CPU
> cycles when the named process only is taking up > 10% CPU..
> and less than 10% in RAM...
> Just looking for any suggested tests..

I'm not an expert on this, so take whatever I say here with a grain of salt 
:D

You could run some dnsperf / resperf 
(<http://www.nominum.com/resources/measurement-tools>) ?
Do some runs without query logging, then some with it enabled. Do the tests 
in the same way. A rather static way of doing it would be to run resperf 
against something you _know_ you have in your cache (like, looking up 
"localhost" or the reverse of 127.0.0.1 or whatever). Do it from the same 
server, or from another server in the same subnet, so you avoid network 
performance.
resperf has a report tool which can easily make some nice graphs for you, 
showing when BIND starts to struggle with sending the replies, and another 
graph to tell you the latency / delay in replies.
This should give you some numbers, to see how much query logging would 
impact you.

Regards
Eivind Olsen