On 8/2/2017 7:10 AM, JORDI PALET MARTINEZ wrote: > I asked a few emails ago, what is our target with any change in the main SSID, for example: > 1) Try real world (real world has IPv4 behind NAT), using for example CLAT, so the WAN can be IPv6-only > 2) Try IPv6-only in the LAN, which we know many things will fail if they use literal IPv4 addresses > 3) Something else? > 4) Marketing “hey we are the IETF, we use IPv6-only, everything works” which can never be true, because if anyone has an IPv4 only device or app, will not work … It is better marketing to make sure that CPEs and OS support CLAT so nothing gets broken? The actual goal of the main IETF network? From the beginning, it has been "provide the best possible Internet connectivity to IETF participants." Which for us today means dual stack, native IPv4 and IPv6, and high bandwidth. That message was repeated quite loudly on the list. The rough consensus is that any experiment should be opt-in, on a separate network. Would that change some day? Yes, very possibly, because of evolving infrastructure and costs. We could see a situation where providing native IPv4 connectivity is becoming too hard, or too expensive. If that happened today, switching to IPv6 plus CLAT would make sense. I assume that this is a long way from happening, but I have not seen the break down of the costs. As for the marketing statement today, it is simple. We do use IPv6 in a dual stack configuration. Dual stack has been the official IPv6 transition strategy for years, so there is really nothing wrong in saying that the IETF uses it. -- Christian Huitema