[ntp:questions] Something which may help improve delays
"terje.mathisen at tmsw.no" at ntp.org
Wed May 23 06:43:05 UTC 2012
Rick Jones wrote:
> I did a very cursory check to see if this was already brought-up but
> may have missed it. Folks interested in addressing highly variable
> delays in networks and such may find the "codel" AQM interesting. It
> seeks to keep the time packets spend in queues down as part of
> addressing "buffer bloat."
> It won't eliminate variable delays and jitter but if it and/or things
> like it get more widely deployed it may constrain them a bit.
I bought a NetGear 3800 a couple of weeks ago in order to run the
CeroWrt test release which is used to verify codel.
So far it works well here, but I have a 50/50 fiber connection
(delivering at least 45-48 Mbits in both directions) so I don't see much
buffering in my end of the network.
PS. I attended a presentation by Jim Gettys, the BufferBloat Man, in
April: The story about how van Jacobsen could not get his new AQM paper
accepted to any refereed conference is both sad and funny:
The paper was dismissed with the comment that "it has to be wrong" and
"the author would do well to read the definitive paper by van Jacobsen". :-)
- <Terje.Mathisen at tmsw.no>
"almost all programming can be viewed as an exercise in caching"
More information about the questions