[Commotion-dev] experience with QoS on OpenWRT?

Ben West ben at gowasabi.net
Mon Sep 22 12:18:32 EDT 2014


To respond to a ping from Dan, I've really not had the opportunity to
exercise the standard qos-scripts package from OpenWRT on links in excess
of 15Mbits on any WasabiNet nodes (which are all ar71xx / mips based).
Primarily because I've no wired uplink yet substantially in excess of
~20Mbits down.  Also, the wireless layers themselves tend to limit rates
before QoS would kick in anyway.

My usual application of qos-scripts is to clamp speeds to single-digit
Mbits range, i.e. to reduce load on the backhaul mesh.  This probably
doesn't exercise nodes nodes enough to make them either CPU or memory bound.

But yes, I would be interested if anyone can share experience on whether
alternate QoS implementations
(i.e. sqm) could perform better on mips-based APs like a TL-WDR7500.


On Sat, Sep 20, 2014 at 2:32 AM, Dave Taht <dave.taht at gmail.com> wrote:

>
> On Sep 19, 2014 12:20 PM, "Dan Staples" <danstaples at opentechinstitute.org>
> wrote:
> >
> > Has anyone done much with the Quality of Service options in OpenWRT? I'm
> > particularly curious if anyone has seen better or worse performance
> > depending on a router's hardware specs (RAM and such). We're including
> > some documentation on doing QoS but we want to make sure to note any
> > problems that users may encounter depending on their router.
>
> I have done a bit here and there. The QoS scripts for openwrt are a bit
> flawed, so there is this thing we have created called sqm which is an
> openwrt package in the ceropackages-3.10 repo on github.
>
> http://snapon.lab.bufferbloat.net/~cero2/jimreisert/results.html
>
> Ram is generally not a problem.
>
> Cpu is a problem above 50 mbit on ingress on the ancient mips platforms.
> htb or HFSC  basically bottleneck  on interrupts. Based on that problem we
> have been developing a new rate limiter, contact me off list if you want
> to try it...
>
> x86 hardware can do htb to the hundreds of mbits. I don't have data yet o
> the newer arm boxes.
>
> Most of the need for inbound rate limiting declines as you go higher than
> 80
> Mbits. It is most needed the slower you need to go on outbound and inbound.
>
> http://burntchrome.blogspot.gr/2014_08_01_archive.html
>
> >
> > 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
>
> _______________________________________________
> 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
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.chambana.net/pipermail/commotion-dev/attachments/20140922/515e60f6/attachment.html>


More information about the Commotion-dev mailing list