Mailing List Archive

For the record, folks, I long ago just sent Abraham's mail into spam,
for me, and just the backscatter from that conversation this week was
bad for my blood pressure. I know he sincerely believes in his
concept(s), and I don't, nor am I going to waste further time arguing
with him.

It's hard enough to stay rational facing the enormous difficulties of
merely shifting 240/4 from reserved to unicast in the iana registry.

Anyway, nice subject change!

<rant>

On Sat, Jan 13, 2024 at 11:59?AM Giorgio Bonfiglio via NANOG
<nanog@nanog.org> wrote:
>
> [.I’m gonna change subject to this sub thread since it makes sense… as opposed to the other 15 subject line changes this thread has seen so far]
>
> First, there is very little cost to any large tech company to state a date that they expect to turn off IPv4 for certain services. "To get our free xyz service after January 1, 2028, you'll need to be on a provider that supports IPv6". The tech companies can then push out that deadline if they don't see enough adoption as the deadline approaches.

2128 seems more likely. And it would be funnier if a bunch of
companies did announce that as a date.

>
>
> You’re suggesting they should bluff, essentially - this is great!

I kind of view the RDOF debacle as of this nature - state a major
policy goal, promise enormous funding for it, watch companies
pre-invest to get on top of the requirement - then pull the rug out
from everyone by inventing insane requirements post-hoc and not
distributing the funds.

Suggesting anyone do the same to support IPv6 kind of rubs me the
wrong way in the same way.

> There are, of course, risks related to consumers switching to other alternatives prior to the date and also various other reputation and legal risks.

In pre-history, and I forget when it was, someone put an enormous
stash of pr0n out there on ipv6 only. Didn't help.

>
> Exactly, which is why this won’t happen until the $$ saving due to abandoning v4 outweighs the risks above.

2128. Among other things, have you seen the k8 mess lately?

> I guess if one was to abstract the above out at a very high level it would be to say that about the easiest way that I can see to further accelerate IPv6 adoption is to either start to provide certain desirable services only over IPv6 or at least threaten to do so.
>
>
> Did you consider a legislative / government level incentive? The EU just forced all mobile devices sold in the area to have an USB-C port, why not enforcing a requirement for IPv6 on all broadband and mobile connections?

I certainly now think some government involvement is necessary.

I would have liked an ipv6 requirement to emerge for BEAD funding, but
most of the lobbying dollars have merely been spent on getting fiber
thoroughly subsidized, instead of routers that run modern software,
have bufferbloat fixes and ipv6. $.30/house verses $30k/mile.

If only a few states distributing that 70B tied an IPv6 requirement
onto it it would help. Please bug your local broadband offices? In
doing up the recent bufferbloat.net FCC filing (which we have
discussed here), we formed a mailing list that several prominent
figures joined. Example:

https://lists.bufferbloat.net/pipermail/nnagain/2023-November/000345.html

It was news to that layer that A) we were out of IPv4 and B) the
government had 10 /8s they were not using and C) 1/16 of the original
internet was still unallocated due to internal infighting amongst the
calcified institutions of the internet and D) IPv6 was not being
deployed by default, still, in many places, in the USA, and other
nations "were ahead".

Some of the arguments we made towards not being able to cross the
"digital divide" without better numbering (be it ipv4 or ipv6) seemed
to resonate, but it is still early days, more people should reach out
to their state broadband offices on these issues, if I can stress that
again!

It was news to several legal interns in broadband regulation that I
have met recently, exactly what a packet was. There is a long way to
go here towards coherent regulation. Perhaps by 2228?

While I was glad to have reached across the net/gov isle last year to
have made this attempt, I would much rather go back to shipping the
next million subscribers for libreqos than fight much more at this
level.

If everyone here made one phone call to their local broadband office,
or (better) showed up at one of the rather nice parties the fiber
industry is perpetually throwing for them - it would help more than
all the discussion on this mailing list to date to push things
forward.




> Or at least a mechanism to clearly identify dual stack broadband offerings vs v4 only?

The FCC broadband labels thing is still mostly a joke. But feedback is
still being solicited.

> US is also mandating a v6 support requirement for gov-contracts - why not extending?
>
> Ultimately - the commercial world can create incentives, but lawmakers can enforce.

I think even the plausible threat of regulation would create
incentives. I also think the plausible threat of one of the agencies
that are supposedly being stewards of the internet to start making
plans for 240/4 to become public would spur the googles and amazons of
the world to get off of it and more onto v6.

> The Googles of the world just happen to be in the best position to do just that and may have a financial motivation to do so (if they can do so without negatively impacting their bottom line).

I do not agree that anyone has an incentive to shift their traffic
exclusively to ipv6, I mostly point out that ipv6 along the edge needs
to deploy somehow.

>
>
> On Sat, Jan 13, 2024, 12:42?AM Giorgio Bonfiglio <me@grg.pw> wrote:
>>
>>
>> 2) Assume that Google decided that they would no longer support IPv4 for any of their services at a specific date a couple of years in the future. […] I really expect something like this to be the next part of the end game for IPv4.
>>
>>
>> It’s never gonna happen … why would Google, or any other internet property, launch something which artificially cuts the potential revenue pool to IPv6-ready customers?
>>
>> I’m with you it would be amazing and a strong driver, but it’s just not in the realm of possibility…



--
40 years of net history, a couple songs:
https://www.youtube.com/watch?v=D9RGX6QFm5E
Dave Täht CSO, LibreQos