Re: [v6ops] Focused discussion: draft-ietf-v6ops-unique-ipv6-prefix-per-host

["Eric Vyncke (evyncke)" <[e-mail]>]

Belated comments on this useful draft...

I am confused by the wording of 'IPv6 only experience even if IPv4 support is present'

It appears to me that the security among users only relies on the AP in isolation mode (or split-horizon)... This makes the I-D mainly relevant to your use case :-(

It is not clear whether the UE will always receive the same /64 even when 'roaming' within the community WiFi as it would have privacy implications.

Section 4.3.2, do we have data backing that most OS still do DNS request over IPv4?

Section 5, I must admit that I do not understand why the GW has to send periodic unsolicited (unicast I hope to get the WiFi ACK) RA... Why not wait for the RS by the UE?

Section 5, when the GW 'pings' the UE, does it use ICMP? Or plain NS ? Can the AP leverage any WiFi polling?

A big  advantage  (which is not described) is that there will be a minimum of NDP multicast traffic which is a pain over WiFI cfr draft-vyncke-6man-mcast-not-efficient-01

Cosmetic:

  *   ASCII art topology would increase the readability of section 3 or move the ASCII art from section 4 in section 3 or remove section 3 or ...
  *   Any chance to replace the HTTP captive portal by a more secure HTTPS captive portal (and I know the issue with certificates are not easy). Or remove it => too specific

I also agree with:

  *   Lorenzo & Tom Herbert: the scope is much broader => the I-D should be made less specific
  *   David: this has nothing to do with AERO :-)

Finally, I cannot refrain from thinking that there are several I-D (including this one) attempting to fix NDP...

Hope this helps

-éric