Hacker News Reader

That tests ICMP echo requests not name resolution.

That is what original test was testing.

What? It says

>Our test was very simple and we performed 70 DNS lookups

That's not ICMP.

Well, the article says so, but they actually measured icmp response. All major domains are in DNS cache anyway. If you look at Yandex result you see it performed very poorly. Why does it resolve popular domains so slow? The reason is it has one server and it is located in Moscow.

So i would say icmp is good proxy to actual performance.

It would be a measure of ping + cache latency. That's still different than just ping.

And is your point that the tested sites aren't popular in Russia?

>for different popular domains (google, facebook, twitter, gmail, etc)

google.com facebook.com and twitter.com are all in the top 50 sites in Russia[1]. And gmail.com is number 6000 globally[2], so it's unpopular everywhere, not just Russia. Popular sites in Russia tend to be popular elsewhere, and vice versa.

[1] https://www.similarweb.com/top-websites/russian-federation

[2] https://www.alexa.com/siteinfo/gmail.com

all these sites are popular in Russia, so we can presume cache latency is a constant on all tested servers. What left is a ping, packet response time.

Reply on news.ycombinator.com