[ppml] IPv6 Assignment Guidelines, Straw Man #2

[David Williamson]

On Mon, Aug 20, 2007 at 07:18:06PM +0100, michael.dillon at bt.com wrote:
> > Using a common metric as a method of insuring reasonable 
> > efficiency doesn't strike me as burdensome. 
> 
> Since when is efficiency a goal in IPv6. I always thought that IPv6 was
> designed so that efficiency was no longer an issue. Instead, IPv6
> enables network designers to build their networks the way they want to
> without worrying about wasting IP addresses.

You seem to have missed the word "reasonable".  Regardless of gigantic
the space is (and it really really is huge), we could easily create an
allocation scheme that would run us out of space if we tried.  I sure
don't want to deal with another round of this every again.  Besides, I
don't think an HD ratio of 0.94 for a /47 or shorter is going to be
even vaguely burdensome on a network designer.
 
> IPv6 is not IPv4 with more bits.

Now that's true.  That's also a key part of the deployment problem.  In
theory, IPv6 was going to solve all the world's ills.  It would allow
efficient multi-homing, solve world hunger, ease global warming, etc.
We've managed to back away from those early promises so that we now
have a bigger address field and incompatible headers.

We'd have been better off just making v4 with more bits, at this point.
(Think how easy a 6-to-4/4-to-6 NAT would be if this was true....)

Anyway, I'm fairly certain we're going to have to agree to disagree.  I
don't see anything in Leo's current proposal that seems excessively
burdensome, especially given the number of customers that most ISPs are
likely to see with legitimate need for more than a /48 of space.

-David