Re: [Cerowrt-devel] [Bloat] tp-link request for SQM

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

On Fri, Dec 3, 2021 at 4:44 AM Kenneth Porter <shiva@sewingwitch.com> wrote:
>
> --On Friday, December 03, 2021 12:12 PM +0000 "Wheelock, Ian"
> <ian.wheelock@commscope.com> wrote:
>
> > 280Mbps service with Comcast is likely the 300/10 package offering… In
> > that case the US is limited to 10Mbps
> >
> > I understand there might be some issue if 280Mbps was being processed –
> > but in the US direction, we are not talking >>100Mbps – its about
> > 10Mbps US, I would have thought running cake on 10Mbps US cake would not
> > have triggered a 50% loss in performance even on this platform. Now if
> > cake is applied to both inbound and outbound traffic then having to deal
> > with ~280Mbps might be tough. In the case of DOCSIS AQM, PIE runs in the
> > GW only on outbound traffic.
>
> That's indeed the package. I was seeing 270-280 down and 12 up.
>
> "Good" tests without cake:
> <https://www.waveform.com/tools/bufferbloat?test-id=22cc9c3e-d3bd-4893-8b68-aba088e199e1>
> <https://www.waveform.com/tools/bufferbloat?test-id=90bfc919-0ccc-47d6-88e6-dee10a4f1b16>

I'm confused, was cake on on the upload? The whole "upload under load
+0" thing I am finding a bit worrisome,
certainly at higher rates with fq on it's going to be close to 0, but
in actually finding one ONT with a mere 5ms
buffer (so far), I am fearing that a solution entering the field is
really short buffers.

Could you take a packet capture of a test without cake?

> It got slammed down to 49 here:
> <https://www.waveform.com/tools/bufferbloat?test-id=a670b084-0e90-434b-aedb-e82b7ea7f7ab>

Ouch. And your uplink went to heck, also. The other worrisome thing
has been powersave and interacting with the
request/grant processes badly.

>
> I disabled cake and it slowly recovered:
> <https://www.waveform.com/tools/bufferbloat?test-id=eb5e874a-b5a4-436a-bb9b-b3a90ae14930>
> <https://www.waveform.com/tools/bufferbloat?test-id=6afcc17b-8d6a-47ee-9a78-ad5b65b74205>
>
> (I wish the tests had a timestamp so I could be sure I'm ordering them
> correctly.)
>
>
> _______________________________________________
> Bloat mailing list
> Bloat@lists.bufferbloat.net
> https://lists.bufferbloat.net/listinfo/bloat



-- 
I tried to build a better future, a few times:
https://wayforward.archive.org/?site=https%3A%2F%2Fwww.icei.org

Dave Täht CEO, TekLibre, LLC