First of all we are talking about several billion more addresses.
Second, you're correct, the NAT kludge has allowed us to delay
IPv6, i.e. simulate global connectivity some of the time. But
it is hardly a strategy for the next hundred years.
IPv6 was designed to help address aggregation, i.e. at least to
start from a point not worse than CIDR. But it was a conscious
choice not to try to invent a new routing model at the same
time. We know that. We have to solve that, *and* we need the
several billion addresses too.
Eric Rosen wrote:
Brian> NAT has simply pushed us back to the pre-1978 situation.
On the contrary, NAT has allowed us to maintain global connectivity without
requiring every system to have a globally unique address. NAT is what has
prevented us from returning to the pre-1978 situation.
That's not to say it wouldn't be better to have a million more globally
unique addresses. Sure it would, unless that would stress out the routing
system unduly. If adding a million more globally unique addresses will
stress out the routing system, then one might argue that a solution which
provides the addresses but doesn't change the routing system isn't really
deployable, and hence doesn't really solve the addressing problem. I think
this is the point that Noel keeps trying to drive home, and I'm not sure I
understand what the answer is supposed to be.