Re: [v6ops] draft-ietf-v6ops-ipv6-roaming-analysis WGLC

[Ross Chandler <[e-mail]>]

On 29 Jul 2014, at 06:10, Mikael Abrahamsson <[e-mail]> wrote:

> On Tue, 29 Jul 2014, GangChen wrote:
> 
>> Only the value IPv4v6 is allowed for ext_pdp. If HLR doesn't support
>> IPv4v6 setting(that is the most case pre-R9), in other words, it can't
>> set ext_pdp. It may not able to support all cases.
> 
> Then I guess it's a matter of implementation. The HLR I was exposed to, did for 2G/3G show to the user that IPv4 could have ext_pdp_type IPv6, and this enabled IPv4 and IPv4v6 to work with this setting. If UE asked for IPv6 only PDP context that didn’t work, so we had to add a profile that said "IPv6" without ext_pdp_context_type.

We have to define two allowed PDP Types for the same APN, so that legacy IPv4 PDPs and IPv6 PDPs for 464xlat can both be requested for the same APN.

e.g. 

<hgppp:pdpcp=240;
HLR PACKET DATA PROTOCOL CONTEXT PROFILE DATA
PDPCP  APNID  EQOSID  VPAA  PDPCH    PDPTY  PDPID EPDPIND
240       80     1    NO             IPV4   25    NO
          76     1    NO             IPV4   26    NO
          74     1    NO             IPV4   31    NO
          73     1    NO             IPV4   32    NO  <-- testdata APN V4 
          73     1    NO             IPV6   50    NO  <-- testdata APN V6

If the Android UE APN Protocol is set to “IPv4/IPv6” with this it sets up parallel bearers to the GGSN/PGW.

I agree with Mikael that it might be worth having a section noting this behaviour. An operator doing 464xlat could have UEs connecting with dual-stack done over separate bearers when they’d prefer it to either be single bearer with legacy IPv4 or IPv6 with 464xlat.  This applies in the home and roaming case.

Ross