[Rpm] [Bloat] [Starlink] [LibreQoS] On FiWi

dan dandenson at gmail.com
Wed Mar 15 13:59:10 EDT 2023


On Wed, Mar 15, 2023 at 11:53 AM Dave Taht <dave.taht at gmail.com> wrote:

> On Wed, Mar 15, 2023 at 10:49 AM rjmcmahon via Rpm
> <rpm at lists.bufferbloat.net> wrote:
> >
> > Agreed, AQM is like an emergency brake. Go ahead and keep it but hope to
> > never need to use it.
>
> Tee-hee, flow queuing is like having a 1024 lanes that can be used for
> everything from pedestrians, to bicycles, to trucks and trains. I
> would settle for FQ everywhere over AQM.
>
> This has been a very fun conversation and I am struggling to keep up.
>
> I have sometimes thought that LiFi (https://lifi.co/) would suddenly
> come out of the woodwork,
> and we would be networking over that through the household.
>
> I'd rather say it's a traffic cop and has value in essentially any
network.  Keeping the costs down on end user hardware is fundamental, and
those devices will behave however they want (ie badly).  AQM is the
'roundabout' that keeps things flowing but each thing at an appropriate
rate so it works well.  There will *never be infinite bandwidth or even
enough that no services saturate it.   Even a very small town with everyone
on a 1G turns into 20Tb of necessary capacity to avoid the usefulness of
AQM.  When likely 20Gb is sufficient.

There has to be something that addresses the car going 180MPH on the
freeway.  That car requires everyone else to pull off the road to avoid
disaster in the same way that data chews up a fifo buffer and wrecks the
rest.  AQM is the solution now, and more evolved AQM is most likely the
answer for many many years to come.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.bufferbloat.net/pipermail/rpm/attachments/20230315/9aa63a01/attachment-0001.html>


More information about the Rpm mailing list