You can use https://cloudflare-test.judge.sh/ to check from which Cloudflare nodes the sites are actually served.

In my experience, sites on Enterprise plan are always loading from the closest node, which is never used by sites on Free plan.

What’s funny is that for me in NYC metro area, all of the Free and Enterprise ones come from the closest location to me (EWR), but the Pro/Business ones are a wide mix, some EWR, some ORD, some YUL. The ones on the Toronto hosted pop are by far the longest ping time, yet the owners of said accounts seemingly paid for the privilege of slower response times. What am I missing?

> ...so I was assuming that would carry over to their caches.

I see mixed results with Cloudflare (pro plan): Sometimes the load times are as low as 50ms, the other times as high as 250ms; both for multiple hours on end, even when I am connecting to the same co-location.

I gave AWS Cloudfront a try as well, and to my surprise their load times held consistently at less than 40ms throughout. It is another thing that, quite unfortunately, for my workloads, Cloudfront turns out to be very expensive.

I get 8ms even to my WiFi router, how can you have such a low response time?

My 1.1.1.1 time is 20ms (Romania).

> I get 8ms even to my WiFi router, how can you have such a low response time?

That’s almost assuredly a sign of radio congestion on the channel you are on causing retransmits. It could also be a poor quality AP or you could have to many clients on your AP, but most likely it’s the first one (in band interference causing lots of retransmits) based on experience.

Yeah, I don't get it either. Even wired I don't think I've ever seen a ping time of under 10ms to anything on the broader Internet here in the UK and I have both FTTC and cable connections. POPs seem to be weird in the UK generally though, I often get geolocated to somewhere 100+ miles away.

No idea frate, though I use a wired desktop for the test, so wifi latency isn't a consideration.

I have 10 ms to 1.1.1.1 in RO/RDS on wired. I blame their idiotic pppoe.

I'm getting single digits from Finland:

    $ ping 1.1.1.1
    PING 1.1.1.1 (1.1.1.1): 56 data bytes
    64 bytes from 1.1.1.1: icmp_seq=0 ttl=58 time=3.766 ms
    64 bytes from 1.1.1.1: icmp_seq=1 ttl=58 time=4.120 ms
    64 bytes from 1.1.1.1: icmp_seq=2 ttl=58 time=4.636 ms
    64 bytes from 1.1.1.1: icmp_seq=3 ttl=58 time=3.587 ms
    64 bytes from 1.1.1.1: icmp_seq=4 ttl=58 time=5.370 ms
    64 bytes from 1.1.1.1: icmp_seq=5 ttl=58 time=3.286 ms
    64 bytes from 1.1.1.1: icmp_seq=6 ttl=58 time=4.084 ms
    64 bytes from 1.1.1.1: icmp_seq=7 ttl=58 time=3.766 ms
    64 bytes from 1.1.1.1: icmp_seq=8 ttl=58 time=3.548 ms
    64 bytes from 1.1.1.1: icmp_seq=9 ttl=58 time=6.760 ms
    64 bytes from 1.1.1.1: icmp_seq=10 ttl=58 time=3.667 ms
    ^C
    --- 1.1.1.1 ping statistics ---
    11 packets transmitted, 11 packets received, 0.0% packet loss
    round-trip min/avg/max/stddev = 3.286/4.235/6.760/0.973 ms

And this is on wifi.

I have ~2.2 ms to 1.1.1.1 on RDS, wired as well. I guess it depends on the area and how lucky you are with the equipment they use.

Frankly 10 ms is good enough.

It depends on how close CF's edge is to you, a normal traceroute from my home's router is 7 hops, but if I'm on VPN, it's just 2 hops from VPN exit.

That seems to imply that you are within about 100km of something on cloudflare's edge.

Yes, I have a suspicion they have an edge node in my city.

They publish a list of their POPs: https://www.cloudflare.com/network/

Ah, thanks! Yep, there's one here.

seems like a worthy correction/addition to your top post that you have lower ping to a cf-pop in your city over cables than to inhouse devices over wifi.

He obviously understands that, hence the “I was assuming that would carry over to their caches.”