Re: [v6ops] new draft: draft-chen-v6ops-ipv6-roaming-analysis

2013/7/18, Tore Anderson <tore@fud.no>:
>> http://tools.ietf.org/html/draft-chen-v6ops-ipv6-roaming-analysis.
>
> Section 4.1 states: "Roaming to IPv4-only networks with IPv6 PDP/PDN
> request would fail to get addresses."
>
> In my experience, this is seldom (if ever) the case.
>
> My home network (Network Norway) supports IPv6 PDP context and in
> recent years I've roamed in more European countries than I can enumerate
> from the top of my head, in some Mid-East countries, in the U.S., and in
> Japan. I've made a point out of trying all available PLMNs my phone can
> see in the air, and I know for a fact that most of them are IPv4-only.
> In spite of this I cannot recall last time I had a problem establishing
> IPv6 connectivity when roaming. The way I see it, 3GPP IPv6 roaming is
> one of those things that Just Works.
>
> I'm not an expert on 3GPP network architecture, but as far as I've been
> able to understand, the reason why this work is that the IPv6 gets
> tunnelled back to my home network using GTP, and that the visited
> network just considers the payload as "data" and doesn't really care
> whether it is IPv4 or IPv6.

Thanks for sharing your experience. Yes, you are right.
There is no such issue if the subscriber's traffic get back to the
gateway (e.g. GGSN) in the home network.
The described failure case occurred when the subscriber get an address
from the GGSN in the visited network. And traffic flows would be
transmitted in a local-breakout manner.  3GPP allows such
local-breakout because it has more efficient routing paths. 3GPP also
specified another architecture called "SIPTO". It guarantees the
subscriber would always select the closest GGSN to reside.

Best Regards

Gang
>
> I'm sure there could be exceptions to the above. I've heard several
> people suggest that Japan is particularly problematic in this regard -
> but in my experience it Just Works there, too.

> Tore
> _______________________________________________
> v6ops mailing list
> v6ops@ietf.org
> https://www.ietf.org/mailman/listinfo/v6ops
>