On Sun, 19 Apr 2015, Toke Høiland-Jørgensen wrote:
Why not? They can be a quite useful measure of how competing traffic
performs when bulk flows congest the link. Which for many applications
is more important then the latency experienced by the bulk flow
itself...
I agree, but it's also easy to create a two queue system where ICMP is
sent in the second queue. So for stupid FIFOs, yes, they measure what you
want, but for anything else they don't.
I do agree, though, that other types of measurements are also needed.
Ideally we should have good latency characteristics for *both* competing
traffic and the bulk flows themselves.
My thinking was that instead of using ICMP PING, for instance a simulated
UDP VoIP call could be done, send one packet every 100ms (yes, I know most
VoIP is 20ms or so) and use that. ICMP is not a good measurement, UDP
would be more valid.
--
Mikael Abrahamsson email: [email protected]
_______________________________________________
Bloat mailing list
[email protected]
https://lists.bufferbloat.net/listinfo/bloat
