Re: [v6ops] I-D Action: draft-ietf-v6ops-ipv6-roaming-analysis-02.txt

[GangChen <[e-mail]>]

2014-08-06 15:29 GMT+08:00, Tore Anderson <[e-mail]>:
> (Changing the subject like Ross did.)
>
> * Mikael Abrahamsson
>
>> I've heard reports of platforms in use in southeast asia, I don't
>> remember if it was Korea or some other country, would fail when
>> presented with IPv4v6 capability.
>>
>> The reason why people are not going for IPv4v6 externally is because it
>> breaks things, badly, it seems, due to bugs. So this is why IPv6 only
>> bearer is so appealing, because it's been around for 5-10 years and very
>> few things break with it.
>
> Yep, and to be clear, Telenor Norway does not support the IPv4v6 on
> their IPv6-capable APN (nor does Tele2/Network Norway). Furthermore,
> Telenor only supports IPv6 (not IPv4), while Tele2/NwN supports both
> IPv6 and IPv4.
>
> Which leads me to a point I forgot to mention before, if Telenor had
> used Roaming Protocol = IPv4 as Ross suggested, or if the phone RIL did
> a similar fallback, it would actually ascertain failure, since the APN
> doesn't allow IPv4. If IPv6 fails, one would have to change to one of
> the IPv4-only APNs.
>
>> The scary part isn't enabling IPv6, it's enabling IPv4v6 or capability
>> to do so. When advertised as capability to visiting SGSN it might refuse
>> to bring up any bearer at all. So this is why the draft suggests it
>> might be a good idea to not advertise this to roaming partners.
>
> Yes, the -02 draft makes it much clearer than previous versions that
> this concern is applicable specifically for IPv4v6 and not for IPv6.
>
> In spite of this I see that some providers that are using IPv6 and home
> routing chose to set the Roaming Protocol to IPv4. This is what I don't
> quite understand the reason for. Perhaps due to a misunderstanding that
> the problems IPv4v6 also apply for IPv6, or some other problem I have
> yet to come across?

Setting roaming APN with IPv4 is recommended in local breakout mode.
As described in
http://tools.ietf.org/html/draft-ietf-v6ops-ipv6-roaming-analysis-02#section-5.3,
IPv6-only PDP/PDN can't be failed over to IPv4. In your case, I guess
traffic is home routed. When a Telenor customer roams Tele2/NwN, All
PDP/PDN request and traffic will back to Telenor network. Therefore,
it works. However, it increases the international data roaming fees to
customers, since Telenor has to pay GRX/IPX carrier for the
international traffic transit. In order to reduce the cost, local
breakout is an optimized approach. It will save the charge for GRX/IPX
transit and improve data service performance. In that case, I guess
the Telenor customer has to check if roaming APN set to IPv4-only,
since the visited network may not able to allocate IPv6 to customers.

BRs

Gang

> I don't really know much about what's going on "under the hood" in
> mobile networks to be honest, but with my "dumb user" hat on I can
> conclude that IPv6 roaming seems to work just fine with the way Telenor
> and Tele2/NwN have implemented it. So I'm happy. :-)
>
> Tore
>
> _______________________________________________
> v6ops mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/v6ops
>