Remco van Mook wrote: Note that inappropriate characters in my environment is automatically converted to question marks.
Dear Masataka,
Would you please enlighten us by sharing how this scheme would work ?
I'm saying the scheme should work.
I?m intrigued. Given the current rate of consumption I don?t see how we could possibly not hit the last /8. If you think the single act of mandating NAT will make the rate of IPv4 consumption stop on a dime it would have happened long ago and at the same time would not change the non-technical aspects of the problem at all.
The current rate is the rate without NAT mandated. With mandated NAT, ISPs have difficulty to authorize their address requests unless the amount of the requests reduced by NAT. If NAT is mandated before the final /8 (among classes A, B and C), we have much time to support unicast class E, which further delay the final /8.
Regardless of how exactly you do it, adding NAT (or any other form of complexity) inevitably adds cost. LIRs on the whole are strongly cost-driven, especially in the current financial climate; investing in more kit than strictly required is a no-sell business case.
Considering the so much delayed deployment of IPv6, when we start allocating the final /8, addition of NAT is inevitable, especially because buying already allocated space will cost a lot. Moreover, adding IPv6 costs a lot more than adding NAT, even when IPv6 is not popular and used by few customers, which is partly why IPv6 is not and will not be deployed very quickly.
Just using up more IPv4 is cheaper in the short run than any other alternative, so it will be the preferred way for a lot of people.
And, it is not fair for those people requesting IPv4 in the future.
Can we please have the discussion about what to do with the last /8 at the risk of not ever needing the bit of policy that comes out of it?
The best thing we can do with the last /8 is to prevent the occurrence of it. Masataka Ohta