> How will portable/non-portable address space be effected. Does the mean
>that if I have non-portable address space, I can charge customers for
>address space, and if it is portable, they have to pay $2500 per /24 to
>have it registered to them?
That would not go over well. <G>
<Warning: the following is not entirely serious>
Oh I don't know, a $122m grant for some competition perhaps? Internic currently
assigns (effectively) out of a /8. There are several other /8s around that
are underutilised. Let's say you run one of these and your network could
actually fit in a /10. You renumber (ouch, expensive), then assign the rest
of your /8 (49152 class Cs) in a manner similar to Internic, doing your best
to assign CIDR friendly blocks etc. etc. to applicants. If you sold them
all at $2500 per class C, you'd pocket $122m (enough to pay for any renumbering
and the admin of the exercise). Would they get routed? Well you would be
assigning blocks of all sizes and if your assignment policy was as good as
or better than Internic, why would people want to filter? In fact, if you
undercut Internic, you could perhaps even sell class Bs etc. to tier-1s
who'd have no reason not to use them. If not, the change from $122m can
pay for a pretty decent amount of transit for some proxy aggregation (i.e.
you continue to announce the /8 for a while).
Perhaps better still, pervert the RIPE model. Sell class /16s out of your /8
to competing subregistries, who can then onsell space at a profit (as smaller
bits are more expensive). May be even reinvent pyramid selling: "Look, you
have this /n currently, all you have to do is renumber with our $5000
autorenumbering software into our /n-2 and sell the plan and the other /n-2 to
3 organizations for $5000, and you'll have instantly made $10,000! Then
sell your old /n using this scheme as well to someone with a /n+2 and make
another $5000". Or something.
Why do I feel there is a flaw somewhere....