Re: [v6ops] The need for local-ipv4 socket transition solutions -- NAT64/DNS64 remains insufficient

[Ted Lemon <[e-mail]>]

On Apr 3, 2015, at 4:56 PM, James Woodyatt <[e-mail]> wrote:
> I see myself advocating against permitting the carriers to escape responsibility. As I view it, they snoozed while operating systems people worked diligently to make dual-stack hosts ready for the transition. Now they'd like an easy escape— asking for host operating systems and applications to maintain the transitional state indefinitely, while they make a clean break to green fields now.

I'm sympathetic, but I don't think this helps us.   What we want is for the transition to happen.   Focusing on fairness isn't going to get us anywhere because fairness really doesn't seem to motivate large corporations to spend money unless there's substantial public visibility, and I don't see that happening here.

What I think is a more viable path to v6only is simply that at some point it might become possible to charge a premium for NAT64, or else silently degrade service on NAT64 without entirely breaking it, e.g. by restricting the number of ports any host can get.   This will put v4-only apps and apps with IPv4 literals at a disadvantage, which might motivate them to upgrade.

Another thing that could motivate them to upgrade is if hosting providers start to charge extra for IPv4 transit, because the amount of market share they expect to lose as a result will not be significant enough to worry them.   But I think this will follow the NAT64 translator cost effect, not lead it.

The point being that I think 464XLAT is going to speed the transition, not slow it, and if that sucks for O.S. vendors, well, that's why they call it work...