3 ms·

Interesting; those techs do not seem to be on djb's radar in his undated, and rather outdated, diatribe. https://cr.yp.to/djbdns/ipv6mess.html https://cr.yp.to

Interesting; those techs do not seem to be on djb's radar in his undated, and rather outdated, diatribe. https://cr.yp.to/djbdns/ipv6mess.html https://cr.yp.to/djbdns/ipv6mess.html So if NAT64+DNS64 moderate this, it seems like there's no meaningful barrier to IPv6-only clients. But, NAT64 translators seem as user-hostile as CGNAT. If this is "transition only", at what point can those translators be removed? Is there an exit plan that would present a smooth and clean, IPv6-only experience, once full transition is accomplished? Or will these be an enduring crutch for decades to come?

wmf 2y ago
DJB's rant wasn't even great at the time and technology has advanced since then. Sadly, a bunch of people have gotten a pretty wrong understanding of IPv6 by reading it. And of course DJB has neither updated it nor deleted it. NAT64 is indeed a form of CGNAT, but everyone today is behind NAT44 so it's no worse. And remember that you can do P2P over native IPv6 so you're better off in that way. Networks like T-Mobile that use NAT64 today are seeing >50% native IPv6 traffic and <50% NAT64 traffic. Eventually the NAT64 traffic gets smaller and smaller and when it hits ~1% you could decide to turn it off.