On 21-07-19 16:26, Puneet Sood wrote: > * The experiment was run from Princeton, New Jersey in Northeast US. > The location is in a very well connected part of the world between > network peering points in NYC and Washington DC. You will not see much > difference (due to network latency) between the cloud providers and > the default (local) Do53. Running the experiment from locations which > are further away from cloud providers would provide another > interesting set of data.
Thanks for this Puneet and for stating this at the microphone during the presentation. I have some numbers that support this. At the hackathon at the African Internet Summit in Kampala Uganda one month ago, I participated with a team using RIPE Atlas to compare response time of the locally configured resolvers of RIPE Atlas probes in the Africa region, to the cloud resolvers 1.1.1.1, 8.8.8.8 and 9.9.9.9. On UDP, on the median, local resolvers returned responses 27 times quicker than the fastest cloud resolver: 2ms RTT for the local resolver versus 55ms for the fastest cloud resolver (which was 9.9.9.9). On TCP, local resolvers provide responses at least 6 times quicker. 14ms versus 94ms (again Quad9). We also measured DoT which had +-3 times longer response time than TCP on the cloud DNS providers. There were only very few probes with resolvers other than the quads that supported DoT, but those were also remote given their long round trip times. See also our post/report on: https://labs.ripe.net/Members/willem_toorop/hackathon-africa-internet-summit-2019 _______________________________________________ dns-privacy mailing list [email protected] https://www.ietf.org/mailman/listinfo/dns-privacy
