On Fri, 27 Feb 2015, Daniel Havey wrote:
Like I said, I agree that bufferbloat is only 1 part of a two sided
problem. The queue with respect to each flow can be either two large or
too small (and sometimes just right?). Anyways, I'm just wondering if
there are any measurement studies out there that show the effects of
this problem in a formal way.
There are even videos showing this on youtube, where downloading at the
same time as doing video conferencing means the video conferencing works a
lot worse, and also that loading a web page takes a lot longer (due to DNS
resolving packets, SYNs etc being stuck behind a lot of data packets).
https://www.youtube.com/watch?v=-D-cJNtKwuw for instance.
Jim Gettys and Dave Taht (and others) have done wonderful jobs in
demonstrating that this is indeed a problem. I know gobs of users who know
this is a problem as well, but they don't know how to fix it. People
active in the bufferbloat movement have shown that there are solutions,
and have shown these to work. There are papers and there are videos to
show this. Even the RRUL test tool kit shows this very well if you
understand what "500ms add latency" will mean to your Internet experience.
--
Mikael Abrahamsson email: [email protected]
_______________________________________________
aqm mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/aqm
