> Of course. Perhaps we should call them Unique Global Addresses, (UGAs), ...
all PI and PA would be "UGA" though. ULA-C and ULA-G are more specific. > If I have a neighborhood network and want to peer with adjacent > neighborhoods to make an ad-hoc metro network, I'd be happy to announce my > ULA-G block and accept whatever UGA/ULA-G/ULA-C/ULA routes my neighboring > networks would like to announce (with the exception of those covered by my > block, of course). importantly, you would not accept driveby ad-hoc. i think that if a malware injection van hits your neighborhood with a pringles-can antenna from a nearby hillock, that you should be able to decide based on their RIPE-181 registration data and/or their SBGP or RPKI registration data that you don't know them well enough to admit them into your wireless mesh network. that's not going to happen with straight ULA, which is why ULA-C or ULA-G are necessary for this use case. > I'll be happy to give my hosts U?A addresses so they can communicate with > the rest of the city at WLAN/MAN speeds, as well as PA addresses to > communicate with the rest of the world at Internet speed. note that in brian's best-possible-world scenario, you'd be able to do what you want with PI. however, we're not in that world, and we need the crutch of leading high-end address bits to tell our NATv6 boxes when to act, and to tell our DFZ-path routers when not to act. thus, ULA-G or ULA-C. -------------------------------------------------------------------- IETF IPv6 working group mailing list [email protected] Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
