[arin-ppml] inevitability of NAT?

[Jack Bates]

On 2/7/2011 2:39 PM, Owen DeLong wrote:
> NAT isn't the solution. SPI is the solution. NAT is unnecessary in IPv6.
>

Just because I love arguing with you, Owen.

NPTv6 on 6to4 anycast implementation is currently under way. It has 2 
possible uses (1 I like, 2 I'm iffy on)

1) CGN networks needing 6to4 anycast support can use NPTv6 to NAT the 
2002:bogusCGNIP:: address to ISP prefixes.

2) By using NAT to convert 2002:: addresses to ISP prefixes, you void 
out the need of a return anycast relay in any 6to4 implementation (at 
the cost of having used NAT66).

6to4 is still a temporary transition mechanism. #1 is required to use 
6to4 in CGN environments. #2 is a safeguard mechanism so you don't 
depend on the return anycast relay, but the cost is NAT66 (it's a poor 
mans 6rd to be honest and a hack).

I honestly expect to see it deployed with #1, as CGN is expected to be 
prevalent which voids out the #2 cases.


Jack