Re: [v6ops] Comments on 464xlat-00

[Cameron Byrne <[e-mail]>]

On Fri, Feb 10, 2012 at 8:17 AM, Masanobu Kawashima
<[e-mail]> wrote:
>
> Hi Washam,
>
> If CLAT gets a prefix longer than /64, it does not need to perform ND
> for 464XLAT addresses on WAN interface because 464XLAT addresses are
> assigned to LAN interface or Virtual interface in CLAT. (i.e. 464XLAT
> addresses are not included in WAN Prefix.)
>
> If CLAT gets a prefix shorter than /64, it must perform ND for 464XLAT
> addresses on WAN interface because 464XLAT addresses are assigned to
> WAN interface. (i.e. 464XLAT addresses are included in WAN Prefix.)
>
> Moreover, if CLAT gets a prefix shorter than /96, it can not perform
> CLAT function.
>
> Regards,
> Masanobu
>

Allow me to add a use case that may make this clearer.  In the draft,
we will work on making that language clear.  This is one of the key
benefits of working group feedback.

In the ideal case, DHCP-PD will provide a dedicated /64 that can be
used for sourcing and receiving the RFC 6145 stateless translation
traffic.

In the case of a 3G phone from a 3GPP pre-release 10 network that does
not support DHCP-PD, the CLAT/Phone will receive a /64 from the
network to be used on the WAN 3G interface.  The CLAT function on the
phone will own the lower /96 of /64 and claim those addresses as its
own so that i may source and receive RFC6145 stateless translations.
It is relevant to reserve this address space for data sessions that
originate from the phone as well as the use case of the phone being
used as a wireless router with NDproxy to the 3G network.

This also applies to any landline deployment where there is either
only 1 WAN or LAN /64.  The CLAT must ensure that /96 is owned by the
CLAT for that subnet for RFC 6145 to function properly and not
conflict with the IPv6 addresses of other devices on the subnet.

CB