John C Klensin writes: > So, let's assume that I'm an ISP and (i) I discover that I've > switched to IPv6 to avoid needing to use private addressing in my > core network, (ii) I discover that it is now costing me more to > support IPv4 customers (because they require protocol and address > translation gateways, even with 4-to-6 and similar schemes) than it > does to support native IPv6 customers. (iii) I decide to start > passing those costs along to the IPv4 users, maybe even > disproportionately to get people to migrate. Or suppose that, as an > ISP, I decide I want to save IPv4 addresses for my big-bucks > customers and hence to force those "regular users" to pay the big > bucks to keep using IPv4. Plausible so far. > Now, at least two things impact whether migration occurs at that > stage. One is whether there are still effective options for IPv4 at > a sufficiently low differential price point to justify a switch in > providers. How large that differential would need to be is pretty > much speculation -- far harder than predicting the future of address > space exhaustion. And it is complicated by the question of how much > choice of providers that regular user actually has -- in many areas, > the answer is not a lot of choices. In the areas that make the heaviest use of the Internet, there will be many choices, and the only ISPs able to get away with an IPv4 surcharge will be the last ones to support IPv4. The first one to attempt a surcharge will inevitably lose customers. > The second is whether IPv6 is really good enough to deliver > services (at the applications layer, which is all those "regular > users" care about) that are roughly as good, and as complete as > set, as the IPv4 services. It is there that I think we are in > trouble with regard to hardware, support costs, tutorial > information, etc. There will also be trouble if someone decides to use IPv6 services that were never available in IPv4, and discovers that the rest of the world is still not on IPv6. The interesting thing is that the last part of the world to move to IPv6 will probably be the part that has the most IPv4 addresses ... that is, the United States. So anyone with IPv6 will have trouble dealing with hosts in the United States, and that will not help adoption of IPv6. _______________________________________________ Ietf@xxxxxxxx https://www1.ietf.org/mailman/listinfo/ietf