Le 23 mars 2011 à 08:21, Ole Troan a écrit :

> Remi,
> 
>>>> It seems that a real world problem case is not covered by this
>>>> update: tunneling v6 over v4 when there is a legacy CPE NAT in the way,
>>>> in an ISP-managed way (unlike Teredo). At the moment this is a well
>>>> known but orphaned problem, which will remain with us until the last
>>>> NAT44-only consumer CPE device has gone.
>>>> 
>>>> It's certainly the case that this scenario doesn't quite match
>>>> RFC 4925 and doesn't need to "support all combinations of IP versions
>>>> over one other." However, that is just a matter of charter wordsmithing.
>>>> 
>>>> Nit: there's a reference to NAT-PT; maybe it should be NAT64 these days.
>>> 
>>> that was the problem one set out to solve with RFC5571 (L2TP).
>>> what's missing from that solution?
>> 
>> Nothing that I know for its solution space, but this solution space is only 
>> stateful and hub-an-spoke.
>> In particular, two host in the same customer site communicate in IPv6 via 
>> the L2TP Network server, which is an important functional limitation. 
>> 
>> As you know, as co-author of RFC 5969 on 6rd, a complementary and useful 
>> solution space is that of stateless solutions.
>> In this pace, we all know that 6rd isn't sufficient to rapidly deploy IPv6 
>> where there are legacy CPE's.
>> A solution for this complementary scenario is therefore desirable.
>> As it happens to be also possible, as shown in the 6a44 proposal, a work 
>> item on it has to be identified.
>> 
>> Softwire, where 6rd has been in scope, seems the logical place to finalize 
>> this complementary solution. 
>> (It is also the logical place to finalize the 4rd solution, but this wasn't 
>> your question and is a different discussion)
> 
> ah, OK I see.
> stateless, managed, tunnels to a host behind a non IPv6 capable CPE. I missed 
> the Teredo reference in Brian's mail.
> 
> I don't quite understand how these types of tunnels can be provisioned in a 
> "managed" way though??

It's plug and play in hosts (6a44 relays are the only ones to need management).
- Hosts obtain the necessary parameters by means of a bubble dialogue with 6a44 
relays.
- 6a44 relays are reached at a well-known address and port couple.

We can discuss this in more details in Prague.

Cheers,
RD

 

> 
> cheers,
> Ole


_______________________________________________
Softwires mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/softwires

Reply via email to