Re: [Bismark-devel] Note the following about RED.

[Dave Taht <dave.taht@gmail.com>]

[-- Attachment #1: Type: text/plain, Size: 1719 bytes --]

On Mon, May 23, 2011 at 7:25 AM, Jim Gettys <jg@freedesktop.org> wrote:

> Van Jacobson is quite pessimistic that you can find a RED 93 operating
> point that will work well in the face of Powerboost like variable bandwidth
> features in home broadband environments.  So if you get into trouble with
> it, don't be too surprised.
>


I'm not expecting to find powerboost-like behavior in South Africa. I have
taken quite a few traces at this point that could use some analysis,
however.



> In fact, its use at all in the edge of the network he believes is
> problematic.  There is too little aggregation and too much variability of
> traffic in a home environment for he or Kathie Nichols to have any belief in
> sane behaviour  in RED 93 given RED's problems.
>
>
Ergo our interests in SFB and RED Light and eBDP (which Van pointed us at in
> the first place).
>                            - Jim
>
>
Regrettably SFB as currently implemented, is as yet untested, nor are
reasonable defaults known, RED Light unavailable, and...

The RED part of openwrt's "qos-script" has been shipped on several million
routers, and it's behavior is not well understood. It uses hfsc, SFB, and
RED to do its work.

After analysis, under South African circumstances, RED is mostly not
triggered, the hfsc portion of the script is where most of the action is
taking place.


http://www.bufferbloat.net/issues/171




>
> _______________________________________________
> Bismark-devel mailing list
> Bismark-devel@lists.bufferbloat.net
> https://lists.bufferbloat.net/listinfo/bismark-devel
>



-- 
Dave Täht
SKYPE: davetaht
US Tel: 1-239-829-5608
http://the-edge.blogspot.com

[-- Attachment #2: Type: text/html, Size: 2788 bytes --]