Maintained by: NLnet Labs

[Unbound-users] unbound monitoring

Alexander Clouter
Tue Aug 24 22:22:19 CEST 2010


hugo hugoo <hugobxl at hotmail.com> wrote:
> 
> I plan to use UNBOUND to replace another DNS software.
> 
> My problem is the following one:
> 
> - How can I be sure that UNBOUND will be as good (or better) than the 
> 	DNS software I will replace?
>
Do you currently have something that benchmarks the performance of your 
*current* DNS infrastructure?  If not then that is the problem you need 
to address before worrying about if the replacement (whatever it ends up 
being) is any better.

> I can see in the statistics the number of received queries sans the 
> number of answers but this do not give an idea of the answer time of 
> the whole system.
> 
> I have seen an answer time for the recursive queries but this do not 
> give an idea of the answer time when the cache is used.
> 
None of this matters till you benchmark your existing system.  After 
that it sounds like you probably do not care what the numbers are (as 
otherwise you would be asking a quite different question), just that you 
can say under test conditions unbound performs 120% or 80% as 
'effective' when compared to brand X.

> Can anyone give me a way to verify that UNBOUND has a good answer 
> time?
>
PCAP records DNS traffic queries to your resolvers for periods of an 
hour at random intervals over the month.  Playback those queries and see 
what happens.

Now the 'correct' question is "what tools do you use to benchmark, 
exercise and torture your DNS infrastructure...".

Cheers

-- 
Alexander Clouter
.sigmonster says: Do molecular biologists wear designer genes?