What is going on in Australia?

[George Bonser]

> 
> Hiliarious. "At worst, instead of a webpage, all users will be able to
> view is a blank page.". It's the strangest writeup because it varies
> between what could be viewed as real responses, and then switches to
> random babbling. According to the article, users might experience
> painfully slow browsing on the IPv6 system, and the IPv4 system will
> not work because of address exhaustion. Can't tell the position of the
> article.
> 
> Truman


I will tell you what could very well happen:

The return of 1980's "walled gardens" where a provider uses the entire
32-bit v4 space and continues operating using v4.  Want to provide
service to a user in that "garden"?  Fine, you will need to get an
allocation from their v4 space and probably tunnel that over v6 to your
operation.

Don't get me wrong, I fully support migration to v6 and am working on it
but v6 has already failed as an alternative to v4.  Large providers will
likely find it easier to simply go back to the "walled garden" approach
than to adopt v6.  The simple fact is that there is too much stuff that
is still broken with v6 including applications, 6to4, etc.  "Happy
eyeballs" is the right way to go but that is ultimately up to the
application developers and we aren't seeing any great rush in that
direction.

There is still practically no end user traffic on ipv6.  What user
traffic there is often breaks, usually in the end user's own network,
and provides a horrible user experience.

V6 is a very good case study in exactly how not to do this again.  V6
should have had a "compatibility mode" with v6 or a 128-bit addressing
mode created for v4 (set the "evil" bit in the v4 header to indicate
that there were six additional fields in the header for the rest of the
source/destination IP addresses).  Send a SYN with the "evil" bit set,
if the SYN came back with the "evil" bit also set, then both sides know
that "extended addressing mode" is supported.  If it comes back without
the "evil" bit or doesn't come back at all, fall back to legacy mode.
This would at least have allowed a v4 stack to address the entire v6
space with a minor change that would have had 10 years to propagate to
hardware in parallel with v6 development.