FCC proposes higher speed goals (100/20 Mbps) for USF providers

Mon Jun 6 15:04:21 UTC 2022

On Mon, Jun 6, 2022 at 7:46 AM Mike Hammett <nanog at ics-il.net> wrote:
>
> "I find it sad that so many would argue for never needing anything
> more than we have today."

My principal argument is that we've made a huge mistake with buffering
in general, all fixed now by various RFCs and widely
available source code, and that if we focused on improving the routers
rather than digging more holes in the ground, the internet would
become vastly better, faster, and cheaper - faster.  If somehow I
could wave a wand and get everyone to reflash a junked router
to openwrt/dd-wrt/merlin/etc and configure SQM calls for moah
bandwidth would decrease. If somehow getting those RFCs mandated in
more RFPs, we'd also be making real progress.

Lower latency does not need more bandwidth it needs better bandwidth.

https://www.bitag.org/documents/BITAG_latency_explained.pdf

If further, folk would stop using oversize wifi channels and
co-ordinate spectrum, again with wifi chips that do the right thing
under contention
(example: https://blog.cerowrt.org/post/fq-codel-unifi6/), it would be
a better network

Or you can burn 10s of k per mile to massively overbuild a fiber
network, that doesn't solve the real last mile problem in wifi queue
management.

>
> Few to none are doing that. Upgrades are an organic part of the process. Some places they're hard, but most places they're comparatively easy. Let's stop putting the cart before the horse just to feel good about ourselves. That's too expensive.

>
> "totally fail to provide the same to everyone."
>
> Why should that be desirable?
>
>
> "If we had moved to fibre everywhere then perhaps"

gbit fiber everywhere would actually work pretty well, as very few
pieces of gear can keep up with gbit fiber:

https://forum.openwrt.org/t/so-you-have-500mbps-1gbps-fiber-and-need-a-router-read-this-first/90305

I was really astonished at how few device at a recent conference could
do a gbit in both directions at the same time. None.

> Negative. DOCSIS works well enough. Modern DSL implementations are good enough. Fixed wireless in many cases is good enough. Next gen satellite is good enough.

... With good queue management. Starlink still has lousy queue
management. Got a blog post coming up.... A lot of fixed wireless,
notably ubnt and now mikrotik, have good queue management. Most DSL
doesn't.

>
>
>
> "If you build it they will come."
>
> So then build the hypothetical content that needs this?
>
>
> Gigabit download level service is available to enough (at least in the US) that if such a downstream heavy service were on our doorstep, it would work for most Americans. Once people got tired of being proven that you need such forward-looking downstream capacity in the regulatory world, they just moved to upstream and cried wolf there too. Yes, many services do have mildly inadequate upstream, but certainly not anything to change the regulatory environment over.

I really would, actually, at this depressing point... like regulators,
to mandate that ipv6 be deployed, and rfc7567 at every fast->slow
transition.
It would raise the bar for adaquate internet services over and above
lost cries for more bandwidth.

>
>
> -----
> Mike Hammett
> Intelligent Computing Solutions
> http://www.ics-il.com
>
> Midwest-IX
> http://www.midwest-ix.com
>
> ________________________________
> From: "Brandon Butterworth" <brandon at rd.bbc.co.uk>
> To: "Mike Hammett" <nanog at ics-il.net>
> Cc: "Michael Thomas" <mike at mtcc.com>, nanog at nanog.org
> Sent: Monday, June 6, 2022 9:31:13 AM
> Subject: Re: FCC proposes higher speed goals (100/20 Mbps) for USF providers
>
> On Mon Jun 06, 2022 at 08:06:50AM -0500, Mike Hammett wrote:
> > "So what happens if the Next Big Thing..."
>
> I find it sad that so many would argue for never needing anything
> more than we have today. It's like why did we bother coming out of
> the trees, or the oceans even (yes Apple digital watches are a pretty
> neat idea).
>
> The non fibre installations we have today, while working for some,
> totally fail to provide the same to everyone. While fixing that
> globally should be a priority it should not be done in a manner
> that will require it all doing again in 10 years.
>
> Building in some headroom for growth makes sense, we're not talking
> lots it's only 10x ish to do gigabit ish, so within error margin.
>
> > I see this said a lot, but it doesn't really mean anything. We
> > are sufficiently close to whatever is likely to come that it
> > can come and bandwidths will have to catch up upon its launch.
>
> If we had moved to fibre everywhere then perhaps, but until
> then we face many decades trying to get that done. So if
> something comes up we may be stuck waiting. Stuff always
> comes up.
>
> When I started the BBC streaming we were told not to bother by
> ISPs, the quality was rubbish, the network couldn't handle it and
> never will. I did it anyway and the net grew but it was a long
> slow process with lots of screaming. It'd be nice to not have to
> wait so long next time because people want to deploy more legacy.
>
> > If we're not that close, then it's unrealistic to pre-build
> > capacity for imaginary developments that never come.
>
> If you build it they will come. People are more likely to
> invest in making things if they see a realistic timescale
> to deployment. If they also have to upgrade everyones home
> too they are less likely to bother.
>
> brandon
>

-- 
FQ World Domination pending: https://blog.cerowrt.org/post/state_of_fq_codel/
Dave Täht CEO, TekLibre, LLC