Re: [v6ops] I-D Action: draft-ietf-v6ops-464xlat-02.txt

2012/4/20, Dan Drown <dan-v6ops@drown.org>:
> Quoting Rémi Després <despres.remi@laposte.net>:
>> Apparently, then, your Android host only offers IPv4 to tethered
>> hosts. (Since it is attached to an IPv6-only network, this
>> restriction should obviously be provisional).
>
> The handset can source IPv4 traffic using the CLAT IPv4 address as
> well as NAT44 the IPv4 tethering traffic.
>
>> In my understanding, that is two /128s that should be proxied:
>> 2607:fb90:800:68c::eae6:901 AND 2607:fb90:800:68c:abcd:0:aff:ff01).
>
> Thankfully, one can re-use the interface ip
> (2607:fb90:800:68c::eae6:901 in this case) on both the outside (cell
> network) and inside (wifi tethering) interfaces.  So it's just the
> CLAT /128 that needs proxy ND.

Just wondering to know how could re-use the ip address, which already
used on outside?
http://tools.ietf.org/html/draft-ietf-dhc-pd-exclude give me a hint it
should be excluded in Lan side

>> Note that this need for TWO proxied addresses, which goes beyond
>> what is done in ordinary IPv6 CPEs, would be avoided with my
>> proposal c2 of
>> http://www.ietf.org/mail-archive/web/v6ops/current/msg12660.html:
>> - CLAT outgoing packets would have SRC =
>> 2607:fb90:800:68c::eae6:901, and DST starting with
>> fd00:976a:c305:692e::/96
>> - CLAT incoming packets (to be submitted to the NAT44) would be
>> recognized by their SRCs starting with fd00:976a:c305:692e::/96
>> - No CLAT IPv4 address would appear anywhere on the wire (but
>> 10.255.255.1 can be used as NAT external address if needed by the
>> NAT44 code)

+1

BRs

Gang

> I am constrained by my implementation choices.  Since I used a tun
> interface and userland code, I need a dedicated IPv6 address for CLAT
> (traffic is sent to the tun interface by ipv6 forwarding).  An
> implementation inside the kernel would be able to do what you ask, but
> it would come at a higher maintenance cost.  Perhaps my concerns are
> outside the relm of what a specification should be concerned about,
> though.
> _______________________________________________
> v6ops mailing list
> v6ops@ietf.org
> https://www.ietf.org/mailman/listinfo/v6ops
>