[Commotion-dev] mitigate Bufferbloat?

[Teco Boot]

CoDel (or variants) needs some time to adapt in dynamic environments, as is often the case in wireless. I don’t think this is a problem.
It doesn’t shape traffic, it manages packet dispatcher in case of queuing.
The problem might be that WLAN drivers have their own large fifo queue, so CoDel can’t manage packet dispatcher. Better integration of AQM at OS level and interface driver FIFO transmission queue is needed.

It has less to do with sgwdynspeed. That one is for selecting the border router based on larger up/down rate and lower ETX.

Teco
 
Op 2 dec. 2013, om 17:50 heeft Ben West <ben at gowasabi.net> het volgende geschreven:

> The casual research I'd done already checking out various CoDel implementations, in addition to existing means of rate control, suggest that they all hinge on the ability to define overall up / down bandwidth values for the rate control algorithm to use as constraint.  This works just fine for most wireline situations (e.g. DSL), as the actual line bandwidth is often known to a good degree of precision.
> 
> Not so with wireless, where a given link's actual bandwidth will fluctuate with external factors like noise.  Especially so for an AP talking to multiple clients, where each client will have a particular rate they've negotiated with the AP, and those rates will further fluctuate based on interference and range.  Deploying rate control incorrectly in such situations seems it could be ineffective at best, or even disruptive at worst?
> 
> I hadn't yet found a CoDel recipe (or other rate control scheme) that appeared able to accommodate variable link bandwidth.  Anyone else?  Posibly, one may be able to harness OLSR's new sgwdynspeed plugin for this purpose?
> 
> On Mon, Dec 2, 2013 at 10:17 AM, Dan Staples <danstaples at opentechinstitute.org> wrote:
> We got a recommendation on the IRC today for avoiding bufferbloat on
> OpenWRT: http://blog.lxgr.net/posts/2013/01/28/my-openwrt-setup/. I
> don't have time to dig into this at the moment, but I would be curious
> if anyone else has experience with this.
> 
> Dan
> 
> --
> Dan Staples
> 
> Open Technology Institute
> https://commotionwireless.net
> OpenPGP key: http://disman.tl/pgp.asc
> Fingerprint: 2480 095D 4B16 436F 35AB 7305 F670 74ED BD86 43A9
> _______________________________________________
> Commotion-dev mailing list
> Commotion-dev at lists.chambana.net
> https://lists.chambana.net/mailman/listinfo/commotion-dev
> 
> 
> 
> 
> -- 
> Ben West
> http://gowasabi.net
> ben at gowasabi.net
> 314-246-9434
> _______________________________________________
> Commotion-dev mailing list
> Commotion-dev at lists.chambana.net
> https://lists.chambana.net/mailman/listinfo/commotion-dev