[dns-operations] Evaluating resolver performance

[Marek Vavruša]

Hi Ralf,

On 3 February 2015 at 20:13, Ralf Weber <dns at fl1ger.de> wrote:
> Moin!
>
>> On 03 Feb 2015, at 08:54, Marek Vavruša <marek.vavrusa at nic.cz> wrote:
>>
>> does anyone have any experiences with benchmarking recursors?
>> From what I understand, the DNS Rex is pretty much dead and resperf is
>> the only thing widely used.
> Shameless plug (I work for Nominum). dnsperf/resperf are still used a lot.
> You can download the package at: http://nominum.com/measurement-tools/
>
>> What I don't like is that it leaks messages to Internet instead of
>> faking DNS hierarchy on a local interface, thus making the results
>> unreliable. Is there anything else I'm missing?
> If a resolver test leaks traffic to the Internet is dependant on how you
> set your test up. While using live traffic may be ok for an DNS operator
> to evaluate live performance, it's hard to reproduce that and is highly
> dependant on the traffic the operator sees.
>
> So the way you do repeatable test is to setup authoritative servers for
> the root, TLDs and SLDs in the lab, configure the resolver to use these
> and generate the appropriate dnsperf query files for the things you want
> to test.
>
> So long
> -Ralf
> ---
> Ralf Weber
> e: dns at fl1ger.de

So do you have any ready-made test sets including both query files and
data/config for the authoritatives?
This is where it falls short, I presume each implementor has its own
dataset and it's hard to tell how much it represents the reality.
I'd like to have something that I can download, set up according to
the spec and let the magic happen, but I guess there's no free lunch.

Marek

>
>
>