ICMP deprioritization by ISPs isn’t real
It’s common knowledge that ISPs deprioritize ICMP traffic. This concept has been so ingrained in me that I feel dirty when I issue pings, knowing that I’m not getting a real world view into network distance and latency.
Naturally, one of the first things I wanted to see was this ICMP deprioritization in action.
PingCatcher allows us to easily compare latency measurements between ICMP and TCP packets. The following is ICMP and TCP latency to Google, Apple, and Microsoft. Notice anything strange? ICMP latency is never higher than TCP. What!?!
I thought I must be doing something wrong. I tried across different TCP ports and over longer network distances:
ICMP appears to be equal to or better than TCP across every measurement attempt. These results are completely unexpected. Why would this be the case? Is it because I’m not testing during peak traffic / load times?
I wonder if UDP is any different?
Perhaps my testing methodology is flawed? If you have any idea what’s happening here, I’d love to hear from you.