Hi folks, > Basically, one of the reasons the DNS protocol has been so robust is > because of the caching behavior. It greatly reduces traffic, greatly > speeds up lookups.
Just want to provide some numbers on lookups RTT. On experiment 1800 (tab 1 at https://www.isi.edu/~johnh/PAPERS/Moura18a.pdf, also https://atlas.ripe.net/measurements/10507676/), we had: * ~10,000 atlas probes querying their resolvers for a unique domain name * two auth servers hosted at EC2 in Frankfurt * Median RTT values: * cache miss queries: 61.51ms * cache hit queries: 2.94ms I know this is not representative for all scenarios. It only covers two auth servers on the same location (unicast), and if we'd used anycast, cache miss medians will significantly decrease. But at least we have a concrete number that works for some scenarios. /giovane _______________________________________________ dns-privacy mailing list dns-privacy@ietf.org https://www.ietf.org/mailman/listinfo/dns-privacy
