[atlas] Trying to measure Quad9 latency

On 11/23/2017 02:23 AM, Jared Mauch wrote:
> I also created this measurement, which is a ping one to measure latency to identify what countries receive a poor response.
> https://atlas.ripe.net/measurements/10291137/#!probes

All those measurements will only show part of the story though:
Quad9 seem to do some _extremely_ weird nonsense.
We were measuring ICMP-latency just from different machines on the same
subnet and saw _huge_ differences in the responses:

> rtt min/avg/max/mdev = 15.088/15.202/15.366/0.155 ms
> rtt min/avg/max/mdev = 4.066/4.134/4.265/0.106 ms

Those times are reproducible, a machine will always get the same 15 or
10 or 4 ms ping. So their anycasted 9.9.9.9 seems to internally redirect
queries based on source-IP-hash, and if you're unlucky, they redirect
you to a server at the other end of the continent, that has a latency
that is worse than if you used a DNS in another european country.

What this means for your measurement is that if your probes IP changed
it might suddenly get a response time that is worse by a factor, even if
it was still on the exact same network.
-- 
Michael Meier, Zentrale Systeme
Friedrich-Alexander-Universitaet Erlangen-Nuernberg
Regionales Rechenzentrum Erlangen
Martensstrasse 1, 91058 Erlangen, Germany
Tel.: +49 9131 85-28973, Fax: +49 9131 302941
michael.meier at fau.de
www.rrze.fau.de