[Bloat] Really getting 1G out of ISP?

[Sebastian Moeller]

Hi Stephen,


> On Jun 29, 2021, at 21:48, Stephen Hemminger <stephen at networkplumber.org> wrote:
> 
> On Tue, 22 Jun 2021 22:51:18 +0000
> "Livingood, Jason" <Jason_Livingood at comcast.com> wrote:
> 
>>> It doesn't help that all the local ISP's claim 10Mbit upload even with 1G download. Is this a head end provisioning problem or related to Docsis 3.0 (or later) modems?  
>> 
>> I'll cover this in an upcoming technical paper (mid-July I hope). Depending on the DOCSIS version, CMTS, and cable modems involved you may have no AQM, or buffer controls for the cable modem, or AQM (sort of) on the CMTS and in the cable modem. In the Comcast network you should find AQM in the upstream queue on the cable modems for which we have deployed RDK-B software (XB6 and XB7), while other devices would have buffer controls.
>> 
>> JL
>> 
> 
> Just a short update. The cable modem matters. Updated from Docsis 3.0 modem with bad Intel Puma chipset
> to new model with Docsis 3.1 and Broadcom and things are much more stable.

	Glad that helped, rather sad state of affairs that these devices are still in the field. (I am not fan of forced obsolescence or retiring hardware too early, but these devices are simply barely fit for their purpose).

> 
> As far as AQM, in this setup; fq_codel does much better than the Cake configuration. With fq_codel can
> see 700Mbit download speed. It looks like Cake is using more CPU

	Yepp, it turns out that all the additional features cake brings to the table have some computational cost. At some earlier state in development cake was leaner and meaner than HTB+fq_codel, but that did not seem to hold.


> especialy since the Cake configuration
> is using an ifb ingress queue discipline as well.

	Both cake and HTB+fq_codel require tricks for download shaping, either a veth pair or an IFB (both seem to have a similar cost but IFB is much simpler to set up) or for a wired only shaper, one can instantiate the internet-download shaper as egress shaper on the interface towards the LAN. Last I tested on an ancient single core 750 MHz MIPS, avoiding the IFB got a 5% higher shaper limit, so not nothing, but also not a way to have your router punch one weight class higher?

Best Regards
	Sebastian


> _______________________________________________
> Bloat mailing list
> Bloat at lists.bufferbloat.net
> https://lists.bufferbloat.net/listinfo/bloat