New minimum speed for US broadband connections

Don Fanning don at
Mon May 31 06:00:24 UTC 2021

Wow...  Talk about a topic that will start a hornet's nest between
engineers and management every time if "money were no object".  It's sort
of like airlines asking "what should be the seat width of our cabins?"
You're gonna get some heated responses.

Just jumping into this thread after scanning over the general sentiments, I
see people who are alphageeks and those who look at everything
pragmatically.  Neither is wrong, but I agree with some that it's the wrong
question to be asking.  I've seen customer sentiment in general with cable
ISP's that the upstream is too limiting in this day and age of
Twitch/TikTok streams and heavy media users.  Gotta remember that for the
most part, almost all the internet traffic is TCP and thus has that two-way
charge when it comes to consuming data... the more data you receive, the
faster you need to ACK those packets and if you're upstream is limited, you
won't be delivering a quality product.

But more to the point, this is the time where a business could take
advantage of the opportunities being offered in resource capital to upgrade
their entire physical plant.  The minimum should be the best technology you
can offer for your subscriber base.  If you see your subscriber base
planning on utilizing 5G, then you'll need to consider throwing out the old
and upgrading the last mile.  That will be some cost and effort.  But last
I heard: glass fiber is fairly future-proof.

Wireless is great for addressing some of the legacy "last mile" issues -
but unless you're next to the device, you'll be fighting for sufficient
bandwidth - as we see with 2.4Ghz 802.11 networks simply due to physics
(ie: more bandwidth == less rf range).  And since wireless is a shared
resource in and of itself, it's not the solution by itself.

We shouldn't be blocking any improvement in any part of the network.  IPv6
and 5G could truly be an amazing thing and next-level products could be
brought to market - if the network was there.  We've seen many times that
if the bandwidth is available ( and hence the inception of 'Internet2'
which is proving beneficial for all sorts of science research due to the
size of datasets ) there will be applications - even if it's to allow
people to truly be without "boundaries" or freed from working in a specific
location due to physical presence requirements and are able to work
anywhere if bandwidth is available to accommodate and not be impacting.

This is usually the biggest consideration for every teleworker unless the
location provides other value (ie: production or lifestyle benefits).

You may only use enough home internet to stick to the lowest subscriber
package.  In an urban modern family, that would be the exception case.. not
the normal.

I get reluctance, but with Starlink and Amazon going up to orbit with
satellite constellations, Alphabet and FB doing drone/UAV research, etc...
the reason they're doing those projects and paying for the research of
"how @ scale to do x", because the physical plant network providers like
former Bell and Cable have a track record of not showing the willingness to
extend service out to that .09% of consumers within a region because the
customer lives at the end of a 50-mile dirt road and there are no other
subscribers along the way.

The fact that ADSL hasn't locked up this percentage of people is shameful
on a multitude of levels - and that's not even withstanding the security
reality of the next problem: SCADA and security were to manage the jack
pump at oil wells, you need to use a dialup modem just to flick a switch on
or off... let alone the other problems with running a national
infrastructure over the PSTN.

So the correct answer is a different question:  Which fiber technology will
we use to deliver FTTP/FTTH?

>From discussions I monitor in WISP/FISP space, the companies/isp's that
commit to building that future growth, are either causing subscriber growth
or are being called by real estate developers to drop fiber in new
communities because customers are demanding fast internet with their
freshly new homes.  For older locations, it should be a no-brainer and
especially to the farm where agra-biz impacts logistics and systems - like

On Thu, May 27, 2021 at 5:32 PM Sean Donelan <sean at> wrote:

> What should be the new minimum speed for "broadband" in the U.S.?
> This is the list of past minimum broadband speed definitions by year
> year  speed
> 1999  200 kbps in both directions (this was chosen as faster than
> dialup/ISDN speeds)
> 2000  200 kbps in at least one direction (changed because too many service
> providers had 128 kbps upload)
> 2010   4 mbps down / 1 mbps up
> 2015   25 Mbps down / 3 Mbps up (wired)
>          5 Mbps down / 1 Mbps up (wireless)
> 2021   ??? / ??? (some Senators propose 100/100 mbps)
> Not only in major cities, but also rural areas
> Note, the official broadband definition only means service providers can't
> advertise it as "broadband" or qualify for subsidies; not that they must
> deliver better service.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <>

More information about the NANOG mailing list