Re: [arch-d] ipv4 and ipv6 Coexistence.

[Toerless Eckert <[e-mail]>]

Thanks, Fred, inline

On Tue, Feb 25, 2020 at 10:16:24AM -0800, Fred Baker wrote:
> On Feb 24, 2020, at 2:27 PM, Toerless Eckert <[e-mail]> wrote:
> > 
> > In private / controlled networks, the choices are not only IPv4 vs.
> > IPv6 or their interop, but also (SR-)MPLS and even more so L2 ethernet
> > switching.
> 
> You make some important points, but...

> We have had several countries discuss pulling off the Internet or doing something pretty disruptive, such as unilaterally changing from DNS to something else or imposing government-controlled firewalls. My point to them - and I have spoken with more than one - is that disrupting Internet communications affects the flow of international commerce, and therefore money, into their countries.

Let me answer this further below.

> The same point applies in this case. If a company wants to internally only use <something> as a link layer protocol, that's their choice, but if they step very far away from IPv4 and IPv6, it means that they can't do business with their vendors and their customers - they have to in some way translate or overlay at that boundary. What has been beneficial with Internet technologies since they were invented was that one no longer had to think about global deployments of BBN 1822, X.25, Ethernet, or any of a long list of other things; they became carriers for a common architecture enabling global communications.

I think its dismissive to call private or controlled networks
"link layers".  They include global enterprise, service provider, defense,
federal and other federated networks, manufacturing and the like. 

The Internet already is an overlay of sorts and the heavy lifting is
done below it by those private/controlled network re-using IETF TCP/IP
technologies, and a lot more than "The Internet".  Including the likes
of MPLS, SRv6, DiffServ, heck even IntServ or *yikes* IP multicast. And
parts of the IETF are still treating innovation to improve those type of
networks as tangential or subordinated to comply with policies really only
built just with "The Internet" use-case in mind.

Making future IETF technologies better support where the heavy lifting
is done is what i think would serve the IETF well.

[rant]
About your first point: I think that additional entities like
governments "breaking" the Internet is a completely separate point,
from the one i made and argued above.

The vision of the transparent global Internet was a great vision to
me until 1995 when the Internet population exploded and IMHO the 
model started to fail. Since then, the mayority of TCP/IP hosts
connects to the internet through some type of policy filtering,
and the IETF has architecturally not really acknowledged
that reality or the need for it. How are countries policing the
Internet for their constituencies now any different from enterprises
having done the same for decades for their constituencies ? What
else but closing its eyes to the issue did the IETF do in the last 25
years ?

And now its easy to see how a limited number of regulated
"WWW service conglomerates" will happily claim to solve all policy
requirements at their application/gateway level. Because IETF
networking architecture has failed to evolve the "Internet" model to
something a little more based on reality of human society than the
original Internet garden of Eden model that is still canon.
[/rant]

Cheers
    Toerless