Re: Fwd: I-D Action: draft-ietf-v6ops-64share-07.txt
"... for delegating IPv6 prefixes to a LAN."
""... for delegating IPv6 prefixes to a single LAN interface."
Generally I think this intro text should make it a bit clearer that these methods are only extending the /64
to a single LAN interface.
3.1 Scenario 1: No Global Address on the UE
- missing "d" on destined.
"the UE (e.g. DNS caching that requires global connectivity) and prevent proper Path MTU Discovery
[RFC1981] to occur on the UE providing an IPv6 router function."
Regarding breaking PMTUD, I wonder if it might be worth requiring that the MTU of the LAN interface is made
the same as that of the 3GPP interface, and if it isn't the default for the LAN interface (i.e., 1500 for
ethernet), then the LAN interface RAs have the MTU option set to the size of the 3GPP interface. I think that
would prevent the UE needing to participate in PMTUD.
"The UE copies the prefix 2001:db8:ac10:f002::/64 from the 3GPP interface to the LAN interface,"
I think it would be clearer if this text specified that this copying of the /64 prefix was indicating the
prefix was on-link, by adding the prefix to the Prefix List, as per RFC5942, "IPv6 Subnet Model: The
Relationship between Links and Subnet Prefixes"
3.2 Scenario 2: Global Address Only Assigned to LAN
"The movement of the IPv6 prefix from the 3GPP radio interface to the LAN interface may result in long-lived
data connections being terminated during the transition from a host-only mode to router-and-host mode."
I'd suggest adding something like:
"Connections which are likely to be effected are ones that have been specifically bound to the 3GPP radio interface."
3.3 Scenario 3: A Single Global Address Assigned to 3GPP Radio and LAN Interface
"This method also creates complications for ensuring uniqueness for Privacy Extensions [RFC4941].
Privacy Extensions should be disabled on the 3GPP radio interface while this method is enabled."
I think it would be better to advise that when privacy extensions are in use, a single privacy address is
generated, and then used as the anycast address on both interfaces, and the preferred and valid lifetimes
are synchronised, such that the anycast addresses on both interfaces expires at the exact same moment.
Hmm, actually, there might be another more broader issue to consider (broader than this specific
scenario), regarding preferred and valid lifetimes of the /64. What preferred and valid lifetimes are
supplied in the RAs by the 3G network, and should they be copied to the LAN interface RAs? Another issue
related to this is the synchronised aging of the prefix on the 3GPP and LAN interfaces. In the case of DHCPv6
PD, the preferred and valid lifetimes announced in the LAN RAs are to be synchronised with the aging of the
DHCPv6-PD prefix, so that when the PD prefix expires, the addresses on the hosts on the LAN expire at the
v6ops mailing list