bufferbloat-beating customer shaping via LibreQoS

Mon Sep 19 00:36:19 UTC 2022

Thanks for the shoutout, Norman. Preseem isn’t at 50Gbps in 1U yet, but we
will get there.

I hope more folks listen to Dave, open vs. closed source solutions aside —
AQM makes a shocking amount of difference to quality of experience.

Jeremy

On Sun, Sep 18, 2022 at 2:14 PM Norman Jester <nj at ancientalien.com> wrote:

>
> > On Sep 18, 2022, at 12:25 PM, Dave Taht <dave.taht at gmail.com> wrote:
> >
> > There's been a huge uptake in interest lately in doing better per
> > device and per customer shaping, especially for
> > ISPs, in the libreQoS.io project, which is leveraging the best ideas
> > bufferbloat project members have had over the
> > past decade (cake, bpf, xdp) to push an x86 middlebox well past the
> > 10Gbit barrier, on sub-2k boxes, with really
> > good stats on backlogs, drops, and ecn marks. I've long primarily
> > tried to get fq_codel and cake running on the CPE (most recently
> > mikrotik), and that's been taking too long.
> >
> > I have no idea to what extent members of this list have interest in
> > this, but if you know of a smaller ISP with bad bufferbloat,
> > please pass that link along? It's got ridiculously easier to set up as
> > a vm of late.
> >
> > There is presently a design discussion going on over here:
> >
> > https://github.com/rchac/LibreQoS/issues/57
> >
> > And by mentioning it here, today, I'm mostly asking what other real
> > life use cases we should try to tackle? What backend tools should we
> > try to integrate with?
> >
> > --
> > FQ World Domination pending:
> https://blog.cerowrt.org/post/state_of_fq_codel/
> > Dave Täht CEO, TekLibre, LLC
>
> Take a look at Preseem as the features it has and graphs are great. WISPs
> need this type of system and would show added interest if it has those
> charts and metrics. The integrations are good also. HubSpot integration is
> a plus so we can pull user data out of it and add it to their HubSpot
> profiles.
>
> --
Jeremy Austin
jhaustin at gmail.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.nanog.org/pipermail/nanog/attachments/20220918/38f2d232/attachment.html>