On Feb 25, 2013, at 6:56 PM, Mark Andrews <ma...@isc.org> wrote: > > In message <1d1732d1-ac03-450a-add2-611f2fb1c...@apple.com>, james woodyatt > wri > tes: >> p3. All this pain can be traded away for the reasonably well-understood pain >> of NAT66 and a single ULA prefix with a constant 16-bit subnet identifier spa >> ce, where collisions will be rare and stateless prefix autoconfiguration will >> settle quickly basically every time. I personally don't think that's a good >> trade, but if routed home networks are ever to become the normal setup, then >> I'm very skeptical that my opinion will turn out to be the majority one. > > NAT66 assumes a /48 or else you don't have enough bits to fix the checksums.
You really need to re-read the document. https://tools.ietf.org/html/rfc6296 First, the term "NAT66" refers to stateful Network *Address* Translation, which doesn't update checksums in the prefix at all - it is what we have done in IPv4/IPv4 NAT but done in IPv6/IPv6 junction points, which means replacing the address entirely. If you're talking about stateless prefix translation, which can update the checksum in the prefix, it's Network Prefix Translation, NPT66. The reason we changed the name was that we couldn't get people to distinguish between those two notions; one blocks traffic while the other does not, and there are other ramifications. We obviously still have not succeeded. NPT66 is not limited to /48s. It can update the checksum in the EID if the prefix is longer. And then there's ILNP, which doesn't update the checksum in the translator because it changed the end to end checksum algorithm (the definition of the pseudo-header). > Mark Andrews, ISC > 1 Seymour St., Dundas Valley, NSW 2117, Australia > PHONE: +61 2 9871 4742 INTERNET: ma...@isc.org _______________________________________________ homenet mailing list homenet@ietf.org https://www.ietf.org/mailman/listinfo/homenet