[mdlug] OT: the great IPv6 debate

[Adam Tauno Williams]

On Tue, 2010-04-20 at 16:16 -0400, Dean Durant wrote:
> Hello, I've just finished reading Dan Bernstein's comments on what he
> calls the IPv6 mess. I think his points are valid. 

No, he's wrong; speaking as an admin that has deployed IPv6.  There is
no mess.  It isn't as 'tidy' as a network that uses only one protocol -
but (a) the Internet has been multi-protocol before and (b) the solution
is to relegate IPv4 to second-class status.

> However, someone out there (IETF, IANA, ICANN, I don't know) obviously
> controls those Class E "reserved for future use" addresses. My idea is
> that while it was smart to reserve them, it's not smart to never use
> them. If now is not the time to use them, when will be? Is this an old
> idea, or not workable ? 

Personally, I just don't care, and don't understand why anyone would
care, about resuscitating IPv4.  It is a 20th century protocol in a 21st
century world;  IPv4 is at this point a mountain of hacks,  from
variable subnetting to bolted on multicast to NAT [ugh!].  And still no
real solution for mobile devices.  Let old horses die.

> So I say, whoever controls those addresses should say, OK, we'll give
> you one, or two, or ten, but, by law (pass the law if needed), you
> must use them to route back and forth between IPv4 and IPv6. Give
> financial incentives if necessary. 

No.  There are already at least two techniques for IPv4/IPv6
interoperability;  for tunneling IPv6 over IPv4 and vice versa.  Hey,
tunneling as a concept is built right into IPv6!

> Only give them to the highest level providers. They then can turn
> around and charge a small premium (perhaps) for access to IPv6 in this
> fashion. People can still route directly from IPv6 to IPv6 on their
> own, but Bernstein has pointed out why this will never happen: there
> is no existing IPv6 infrastructure. 

False, False. False.  Check out Hurricane electric.  Get yourself on
IPv6 so you can see - yep, the Internet is there.  An no large IPv6
networks?  Comcast, Verizon, the DoD, several of the large European
ISPs, and a sizable chunk of Asia.

> Bernstein points out that IPv6 was not designed to be an extension to
> IPv4.

And thank goodness.  IPv4 is a withered old crone,  once you start using
IPv6 any fondness you feel for IPv4 will quickly reveal itself as
nothing but nostalgia.

> My idea provides that extension, if my idea makes sense. I'm not
> saying give out all of the class E addresses, but however many as
> makes sense. Maybe just a total of what, 512? 1024? How many in all
> are there? 

There is no point.

> However, I am not a networking professional. Just a curious bystander.
> But, if the U.S. Congress has stated that IPv4 exhaustion is a
> national security issue, and China *is* building an IPv6
> infrastructure, maybe "natting forever" is not going to work. 

NAT is stupid.  All it does it cause interoperability problems and choke
performance.  Just use IPv6.   And the USA *is* building IPv6
infrastructure.  So no cause for alarm.