[ipv6-wg] Implications of NAT/NAT64 and similar

Hi Silvia and list,

Silvia Hagen <silvia.hagen at sunny.ch> writes:

> Transition mechanisms have the attribute of being transitional by
> definition. So they bridge gaps.  That is exactly what NAT was
> originally designed for - bridge a gap until real solution is at hand.
> [...]

yes, but they also have another effect: They make people believe that
the problem is solved, often leading to a situation where these stopgap
measures don't work any longer and the pressure to fix the problem is
*really* bad.

Or, in a different context: As long as the painkillers work, why should
I go to the dentist?

Only when the workaround is more painful than the problem it solves,
then people start to move.  We see that with DS-Lite, which is why I
called it so useful at the RIPE meeting in Amsterdam: To the ISPs and
the content providers, and increasingly to the enterprises too, it is
simply less painful to go server side dual-stack than deal with the
collateral damage caused by IPv4-only on their side and DS-Lite on their
users.

> So let's hope we learned from the NAT issue

Seriously, I doubt that; this behaviour is too deeply entrenched in
peoples minds.  And while we're both making a living out of IPv6, it's
still a niche market for people like us; if people had learned, then
we'd be drowning in project offers.

> We live in a free world and the Internet is VERY diverse.
> What is good for the ones is bad for others.

Fully agree on that.  And it's rather difficult to come up with ideas or
approaches that won't cause significant problems for others.  But that's
exactly why I point out the problems NAT causes to other people.

> I do not believe in IETF or whoever else dictating the market how to
> do it.

Hmm, I guess there are a number of people at the IETF who'd be quite
seriously offended by that statement, but anyway: One of the most
fundamental goals of the Internet and the technologies used there is
interoperability.  I still remember how Compuserve and AOL and MSN and
various others tried to establish their proprietary internetworking
products, and people eventually opted for the Internet largely because
of its interoperability.

The role of the IETF is to ensure that specifications (not standards by
the way, but that's yet another issue:-) support this interoperation.

> I think we should offer different transitional solutions solving
> different issues and let the actors in the market decide, what
> combination works best for them.

Yes, but: If the "market decides" then there's a good chance that people
will sacrifice that interoperability for their own advantage.  There are
some people who benefit from full-blown NAT66, but others will then have
to pay the price (STUN, reduced reliability, increased operational
expenses, extra hardware, ...).

One of the problems with transitional solutions, aside from delaying the
inevitable until it really hurts, and always lasting forever, is that
each increases complexity.  And considering how difficult it is to
recall those solutions when once made available.  Which leads back to
the extension header issue...


Cheers,

    Benedikt

-- 
Benedikt Stockebrand,                   Stepladder IT Training+Consulting
Dipl.-Inform.                           http://www.stepladder-it.com/

          Business Grade IPv6 --- Consulting, Training, Projects

BIVBlog---Benedikt's IT Video Blog: http://www.stepladder-it.com/bivblog/