Thus spake "John C Klensin" <john-ietf(_at_)jck(_dot_)com>
It seems to me that if you, or anyone else, wants to make the
case that the IPv6 address space isn't going to be large enough,
you need to do so explicitly and immediately (five years ago
would have been better, but maybe we know more now). While I,
personally, would have preferred extendable length addresses, it
feels to me that saying "yet" at this stage in the game is just
pointless sniping unless you are willing to argue for calling a
halt at this stage and redesigning.
I see no reason to believe that the address length is insufficient on its
own; however, allocation policies and the way those bits are used may cause
exhaustion and/or a different way of looking at those bits.
Based on the number of hosts connected to the Internet today compared to the
total number of unicast IPv4 addresses, we are nowhere near a shortage.
Even so, pre-CIDR allocation policies resulted in vast amounts of unused
space. Rationing of the remaining space has made NAT a necessary evil just
to keep the beast functional for the masses.
Are we doing the same thing with IPv6? Maybe. At worst, we're not making
any mistakes we can't correct later when we learn more. I'm particularly
suspicious of the decision to use a /64 for each subnet instead of the
original /80 plan, and the allocation of a /3 to the RIRs is mind-boggling.
But I do think we bought ourselves enough headroom to seriously screw up
several times and not need a whole new protocol.
Stephen Sprunk "God does not play dice." --Albert Einstein
CCIE #3723 "God is an inveterate gambler, and He throws the
K5SSS dice at every possible opportunity." --Stephen Hawking