Re: what are the available choices ?

[David R Conrad <[e-mail]>]

Yakov,

>1. efficient address allocation, non-routable Internet
>2. non-efficient address allocation, routable Internet
...
>However, with the second alternative the Internet would
>still be able to grow, as IP addresses would be available from some
>sources (need not be registries) (for more on this see RFC1744 by Geoff
>Huston).

And you would likely have a much more rational allocation methodology
which would eventually lead to more efficient address space usage
("you mean I have to pay the Internet Address Brokers $65,536.00 for a
/16 and my ISP $10,000/year to route it?  Umm.  Give me one of your
/32s").

In such a case, however, I feel there would likely be an explosion of
interest in technologies such as NAT and thus a violation of the
"Internet Architecture" criteria of "End-to-End connectivity" as
described in Stockholm (if I understood that bullet).  Of course,
people would still be using the network in much the same way as they
do now so one might argue "who cares?".

I figure that in an environment where there is a significant charge
for address space and out-of-block routing entries, you would evolve a
public Internet that acts as a connector for a very large number of
private internets which attach through (provider addressed because
it's easy) NAT boxes of one form or another.  In such a world, the
concerns of renumbering essentially vanish and arguments regarding
portability of addresses become meaningless (your entire internet
becomes portable).

>P.S. There is also a third alternative -- non-efficient address
>allocation, non-routable Internet.

Which is, of course, the way current policies tend...

Cheers,
-drc