On Fri, Dec 2, 2011 at 3:22 AM, Lorenzo Colitti <lore...@google.com> wrote:
> On Thu, Dec 1, 2011 at 10:40, Alexandru Petrescu <
> alexandru.petre...@gmail.com> wrote:
>
>> Le 01/12/2011 18:24, Tao Sun a écrit :
>>
>> Hi Jouni, Thanks for the clarification. That is ture GGSN/PGW is not
>>> mandate to do relay. "May" instead of "shall" is used to describe
>>> GGSN/PGW's action. The only mandate is to use RA to obtain /64
>>> prefix.
>>>
>>
>> And that mandate is not sufficient when the UE is actually a Router
>> ("mobile hotspot", "tethering"). For these cases, DHCPv6-PD is
>> suggested by 3GPP specs. Thus, the 3GPP specs assume that a UE-Router
>> (new term for this conversation) runs both RA _and_ DHCP. But 3GPP
>> specs overlook the fact that an UE-Router would't use RA to
>> auto-configure an address (stateless autoconf is only for Hosts).
>
>
> That's not true - RFC 6204 says that an IPv6 CE router can use autoconf to
> create an address. I think it's reasonable to say that if you were to
> implement tethering using DHCPv6 PD, you would follow RFC 6204.
>
>
>> Hence 3GPP specs could consider that _only_ DHCP should should be used
>> by DHCPv6. In this sense, the above only mandate ("use RA to obtain
>> /64 prefix") could be dropped altogether.
>>
>
> See what I mean? We're already talking about not using RAs at all any
> more, and we don't even have a DHCPv6 route option yet...
>
RA still must be used by the host to get the /64 prefix first and establish
the connection towards the gateway. After that, DHCPv6 PD is used if
shorter than /64 prefix is required.
If you worry about the possibility of dropping the use of RA, the draft can
be updated to clearly state that using DHCPv6 for route option shall not
prevent using RA for SLAAC... somthing like that.
I
> _______________________________________________
> mif mailing list
> mif@ietf.org
> https://www.ietf.org/mailman/listinfo/mif
>
>
_______________________________________________
mif mailing list
mif@ietf.org
https://www.ietf.org/mailman/listinfo/mif
