[NNagain] Carr blasts BEAD

Thu Jun 20 17:37:55 EDT 2024

Hi Sebastian,

I think fronthaul & wireless networks will cause consumers to ask more
from the OSPs which will include market payments for OSP upgrades. EVs
seem to be doing this for electric infrastructure.

The fronthaul concentrator probably isn't a switch as room/room isn't
needed.  The energy cost of the TCAMs in switches are too high. I
think the fronthaul concentrator needs a 20 year lifetime and to be
passively cooled. Maybe 25W max.

The thin radios or remote radio heads don't have to be 802.11.  They
basically connect a fiber cable to a wire acting as an antenna or a
patch antenna. These parts are plugable, analogous to an SFP in a data
center.  The buy & upgrade cadence will match the engineering NRE
cadences. 802.11 is the easy choice now per mobile phones have paid
out a large NRE (as game consoles did for Nvidia's GPUs)

This is a multi company or industry wide challenge. A new software
company is likely needed along with system integrators and component
suppliers. And we need lots of installers. Not one person nor even one
company can achieve this.

The NRE is quite large for the new chips. They have to have a
potential to go worldwide otherwise the NRE will never be allocated.

This is not a new story. Ethernet switching origins were FDDI
concentrators though few connected the dots ahead of that evolution.
Now people think ethernet switches have always existed. It took
decades of engineering to get from FDDI to where we are today.

Engineers fundamentally have to lead this as it's hard core engineers
that define the horizon effect. (It's really not much different from
the days when Einstein's father was deploying electricity in Europe. I
suspect those early pioneers had some idea of what they were actually
doing for our societies.)

Bob

On Thu, Jun 20, 2024 at 12:08 AM Sebastian Moeller <moeller0 at gmx.de> wrote:
>
> Hi Bob,
>
>
> > On 20. Jun 2024, at 00:18, Robert McMahon via Nnagain <nnagain at lists.bufferbloat.net> wrote:
> >
> > False dichotomy. Fiber and wireless are the end game.
>
> +1, at least on the fiber part. Sure building out near-universal infrastructure typically has a relative high up-front cost, but tends to increase productivity over longer time frames#. Given that in the long run it is going to be the society that will profit from it, I see no reason why the society should not pay for the deployment... That however IMHO is orthogonal to the "many of the goals could be achieved by better software" argument, as that better software will also help on fiber deployments* and can be useful for bridging gaps until the (slowish) deployment reaches individual areas.
>
> #) Just look at other infrastructure: power lines, wast/waste-water lines, gas pipelines, roads, the POTS network... getting near universal roll-out for each was/is also a costly measure, and still the argument, instead of building roads we should simply subsidise 4x4 off road vehicles for people living in such areas hopefully is an obvious dud. Or rather over emphasises the short term cost over the long term gains. Capitalism in its stock market driven form typically lacks the stamina to really go for the longer term gains and hence often it is left to the tax payer to make such investments happen. And yes, thast comes with quite some bean counting, pencil pushing and delays, but 'better late then n
>
> *) Case in point, still popular GPON typically only offers 2.4/1.2 Gbps segment capacity, with (max) segment size often 32 or 64 subscribers, leaving only 2400/32 = 75 to 2400/64 = 37.5 Mbps of average capacity per user. So during primetime GPON segments will likely see similar disparities between capacity and demand as FWA segments, and will likely desire the same counter measures.
>
>
> > The major fiber installers I know say they no longer pull copper, even in houses. The pull strength and bend ratios of fiber now exceed all others. Fiber is one and done. Fiber cables don't have capillary action when submerged per rainstorms. Signal loss of copper at 100Gb/s is in dB per inch. The sweet spot for optics considering all kpis including power per bit delivered is 100G.
>
> Not doubting that at all, yet I can see that typical end-users will not really be able to fill up such a pipe anytime soon; and I doubt that end-users are more sensitive to $/bit than to total $, so a 100G solution would need to come close to current prices for 1 or 2G5 Gbps ethernet gear and that is including the optics. Not sure 100G switches will become 'affordable' for mere end users any time soon.
>
> > Fiber is less than 0.5 db over 2Km independent of modulation.  Thin as a hair and capacity abundant.
>
> It also is a tad more resistant to RF ingress than copper or wireless networks...
>
> > Wireless is needed for the unleashing of devices and no more. Maybe 29' like smoke detectors.
>
> IMHO WiFi still has a way to come... but also will not go away, so I agree (more grudgingly than to the fiber part).
>
> > Those that figure this out will lead and make for the future and have meaningful impact.
>
> Assuming they can actually make networking gear or are in a position to create requirement lists for big customers of such gear, no?
>
> Regards
>         Sebastian
>
> >
> > Bob
> >
> >
> >
> > On Wed, Jun 19, 2024, 2:22 PM le berger des photons via Nnagain <nnagain at lists.bufferbloat.net> wrote:
> > maybe they'll pay some attention to you in 18 months when the only ISP doing well will be ones that are either owned by the Rothschilds or trade a month of internet for some eggs and a chicken.
> >
> > On Wed, Jun 19, 2024 at 10:57 PM Dave Taht via Nnagain <nnagain at lists.bufferbloat.net> wrote:
> > Some good rhetoric here: https://x.com/BrendanCarrFCC
> >
> > I have been boggling at some of the hurdles required to get this funding. I gave up on BEAD last year, as the goalposts seem to be forever receding and mounted on unicorns dancing on cotton candy clouds. I know there is more progress being made than meets the eye, but to me it's in spite of BEAD, rather than because of it. ("in spite" meaning - you are going to subsidize WHAT??)
> >
> > My pithy comment about the "fiber party".
> >
> > Bead has provided employment to a lot of telephone polishers, political hacks, and paper pushers. I tried to point out that many of the goals could be achieved by better software on wireless networks; the fiber party took it over. And oh! Did they party!
> >
> > https://x.com/mtaht/status/1803510661093392667
> >
> > And for the records, I actually rather liked the ACP program. And LibreQos passed 154 ISPs deploying it yesterday.
> >
> > --
> > https://www.linkedin.com/feed/update/urn:li:activity:7203400057172180992/
> > Donations Drive.
> > Dave Täht CSO, LibreQos
> > _______________________________________________
> > Nnagain mailing list
> > Nnagain at lists.bufferbloat.net
> > https://lists.bufferbloat.net/listinfo/nnagain
> > _______________________________________________
> > Nnagain mailing list
> > Nnagain at lists.bufferbloat.net
> > https://lists.bufferbloat.net/listinfo/nnagain
> > _______________________________________________
> > Nnagain mailing list
> > Nnagain at lists.bufferbloat.net
> > https://lists.bufferbloat.net/listinfo/nnagain
>