[Bloat] Looking for a citation...

[Dave Taht]

One of the things I like about sitting on top of the libreqos deployment is
being able to see the side effects of different ISP bandwidth tiers, and we
can now correlate rtt, drops and tcp retransmits somewhat. At 25/10 cake
manages most of the bandwidth (beautifully). At 100/20 we see very little
ISP drops but do see rtt skyrocket sometimes which is a sure sign to the
ISP to suggest a wifi upgrade of some sort to the customer.

On Sun, Aug 18, 2024 at 9:01 AM David Lang <david at lang.hm> wrote:

> Also, the highest wifi speeds are only achievable with build traffic to a
> single
> client (or with mu-mimo to a small number of clients), it's not that with
> n
> clients, each gets anywhere close to 1/n bandwith. And if you add a single
> slower device to the network, it will eat much more airtime than it's
> bandwidth
> would indicate.
>
> David Lang
>
> On Sun, 18 Aug 2024, Dave Taht via Bloat wrote:
>
> > Date: Sun, 18 Aug 2024 08:47:02 -0700
> > From: Dave Taht via Bloat <bloat at lists.bufferbloat.net>
> > Reply-To: Dave Taht <dave.taht at gmail.com>
> > To: Jan Ceuleers <jan.ceuleers at gmail.com>
> > Cc: bloat at lists.bufferbloat.net
> > Subject: Re: [Bloat] Looking for a citation...
> >
> > Actually I feel that at speeds greater than *50*Mbits, most of the bloat
> > moves to the wifi, but perhaps I should qualify it more, Modern wifi can
> do
> > almost 2gbits a few feet from the AP, but still has a dynamic range of
> > 5Mbit to 2gbit. Interference, contention, range, all factor into when you
> > hit a FIFO "cliff", and stay there.
> >
> > I wish I knew how many commercial APs outside of eero, cisco meraki,
> > gfiber, and starlink have adopted fq_codel. Certainly I am pleased as
> punch
> > at openwrt's adoption. And seeing at least a few fiber folk shipping
> better
> > wifi.
> >
> > Moreso, if only more vendors did a RvRvlatency test like:
> >
> http://flent-newark.bufferbloat.net/~d/Airtime%20based%20queue%20limit%20for%20FQ_CoDel%20in%20wireless%20interface.pdf
> >
> > A hugely mitigating factor is people self adapting to move closer to the
> AP
> > (or mesh), another is most traffic never cracks 20 mbit for very long.
> >
> > I am sad that every coffee shop I frequent save one, has horrible
> > bufferbloat, but it  usually only shows up when you try to do s
> > videoconference.
> >
> >
> >
> > On Sun, Aug 18, 2024 at 8:32 AM Jan Ceuleers via Bloat <
> > bloat at lists.bufferbloat.net> wrote:
> >
> >> On 18/08/2024 11:08, Rich Brown via Bloat wrote:
> >>> In various posts, I have baldly asserted that "above 300-500mbps ISP
> >> links, all the bufferbloat moves into the Wi-Fi."
> >>>
> >>> I am pretty sure that I someone on these lists stated that as fact.
> >>>
> >>> Could I get a link to a discussion that is definitive? Or a statement
> >> that is actually true that I can incorporate into my future posts? Many
> >> thanks.
> >>
> >> Quite evidently there are WiFi access points and clients available whose
> >> speeds exceed 500 Mbit/s, so in order to be able to make such a claim
> >> one would need to know the extent to which those newer WiFi technologies
> >> are not yet deployed.
> >>
> >> _______________________________________________
> >> Bloat mailing list
> >> Bloat at lists.bufferbloat.net
> >> https://lists.bufferbloat.net/listinfo/bloat
> >>
> >
> >
> >_______________________________________________
> Bloat mailing list
> Bloat at lists.bufferbloat.net
> https://lists.bufferbloat.net/listinfo/bloat
>


-- 
Artists/Musician Campout Aug 9-11
https://www.eventbrite.com/e/healing-arts-event-tickets-928910826287
Dave Täht CSO, LibreQos
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.bufferbloat.net/pipermail/bloat/attachments/20240818/514907ab/attachment.html>