> From: Keith Moore <moore@xxxxxxxxxxxxxxxxxxxx>
> Do you actually have a point to make
That depends. Are you still of the opinion that IPv6 will, in our lifetimes,
become ubiquitously deployed, thereby restoring us to a world of transparent
end-end, or do you think we should acknowledge that that's not going to
happen, and start to think about how to design for a permanently mixed
Internet - and actually have that model in mind when doing protocol work?
Because there are clearly a lot of people who don't buy into that (viz. this
recent comment "Now is not the point to invest time fixing the ipv4
internet.").
>> Look what you have done: not only we have more NATv4 than ever, but
>> now we also have NAT46, NAT64, NAT464
> I think you give me far more "credit" than I'm due.
I have to agree. In some sense, NAT became inevitable way back in 1976 (or
so, don't recall the exact date of this), when variable length addresses were
removed from IPv3 (I think it was) in favour of the 32-bit fixed-length
addresses. (The failue to differentiate between interface names and endpoint
names also was a factor, but somewhat tangential.)
The reason is simple: i) 32 bits would eventually be too few for naming
endpoints, and ii) when that happened, since no evolutionary path to a larger
namespace had been defined, naming domains separated by translators were
driven by economics as the cheapest way forward.
Noel
_______________________________________________
Ietf mailing list
Ietf@xxxxxxxx
https://www.ietf.org/mailman/listinfo/ietf