Comment by zamadatix
7 years ago
It would have been interested to see actual delay rather than qualitative results of the nature "<x>% wasn't in cache so this is horrible!". Admins and users don't care if it's in cache, they care what the impact to operations and load time is. https://www.dnsperf.com/dns-speed-benchmark says lookup times for my personal domain results in 20ms-40ms. Ironically the same dns test for 00f.net is taking 100ms-150ms.
99% of apps will gladly trade a 30ms increase in session start (assuming the browser's prefetcher hasn't already beaten them to it) to not have to worry about things taking an hour to change. Not all efficiency is about how technically slick something is.
I just tested 00f.net and got as low numbers as 6ms. Latency is a question about network traffic between the client and the server, and unless you use anycast you will get different latency depending on what place in the world the client and server reside in, and if you use anycast it depend on how good the contracts and spread the anycast network has.
Very true, looks to be hosted out of Europe. The point about 00f.net optimizing for ease of operation vs optimizing milliseconds of performance only holds doubly true with this information though.