Re: [shara] aplusp BOF

[<[e-mail]>]

Hi all,


The current aplusp charter proposed by Ralph says: "The aplusp working group will evaluate the requirements of IPv4 address sharing in the context of dual-stack lite".

"in the context of dual-stack lite" is surely something that must be clarified.

>From my understanding it means three points:

1) A PRR (Port Range Router) function is located on a device where there is also a DS-lite CGN fucntion implemented (this is compulsary). 
This co-location is what I call a SHared Ip Processor (SHIP) in http://tools.ietf.org/html/draft-levis-behave-ipv4-shortage-framework-02 
The PRR function consists in:
a) for outgoing packets, do nothing
b) for incoming packets, from the destination port value find the CPE tunnel endpoint (the IPv6 address)

2) The packets that are PRR-processed use the same encapsulation than the packets that are DS-lite-processed, that is, in the form of: IPv4-into-whatever-into-IPv6-into-whatever.

3) For the CPEs that are federated under a PRR/DS-lite device, I see three possibilities:
-DS-lite only CPE: no NAT, IPv4 in IPv6 encapsulation of some sort
-PR only CPE: NAT, IPv4 in IPv6 encapsulation of some sort
-DS-lite and PR CPE: NAT used only for PR-processed packets, no NAT for DS-lite-processed packets, IPv4 in IPv6 encapsulation of some sort


What do you think? Do you all share this understanding of Ralph's proposal?



Regards,


Pierre




-----Message d'origine-----
De : [e-mail] [mailto:[e-mail]] De la part de [e-mail]
Envoyé : vendredi 9 octobre 2009 08:41
À : [e-mail]; [e-mail]
Cc : [e-mail]
Objet : Re: [shara] aplusp BOF

Ralph,

The DS-Lite essentially is encapsulation+CGN. The most commonly discussed encapsulation method is IPv4-over-IPv6. However, L2TP, PPP, GTP, etc "point-to-point" connections are also possible in addition to plain IPv6. This is mentioned in DS-Lite dradt as well. I believe these other encapsulation means are in the scope of the BOF? 

Best regards,

Teemu
________________________________________
From: [e-mail] [[e-mail]] On Behalf Of ext Ralph Droms [[e-mail]]
Sent: Friday, October 09, 2009 8:07 AM
To: Rémi Després
Cc: [e-mail]
Subject: Re: [shara] aplusp BOF

But generalizations of A+P were argued against in the shara BOF.  A+P
represents a truly significant change to the IPv4 addressing
architecture that makes fundamental changes to the behavior of IP and
applications built on IP when implemented in a host.  By constraining
the problem space to the scenario with well-defined endpoints and a
well-defined forwarding mechanism, the BOF will have an opportunity to
focus on a case where it can be found to be useful.

- Ralph

On Oct 8, 2009, at 6:35 AM 10/8/09, Rémi Després wrote:

> Ralph,
>
> I fully support Randy's view.
>
> Having as A+P *on dual-stack lite* as "main objective" of an aplusp
> WG would be much too restrictive.
>
> Having it as a "minimum objective" would be OK though, because of
> recent progress made on this particular A+P application.
>
> The WG SHOULD be the place to determine which other architectures
> and signaling alternatives address new operational requirements, and
> therefore justify quick progress.
>
> In particular, while dual-stack lite is for IPv6-only ISPs, some
> ISPs will rather deploy IPv6 AND private IPv4 addressing on their
> infrastructures. They should also be able to take advantage of A+P.
> (These advantages are good to have: simple server reachability when
> communicating with IPv4-only remote hosts; restored end-to-end
> transparency for hosts that support A+P.)
>
> Similarly ISPs that only have private IPv4 routing so far, across
> which they can quickly deploy native IPv6 with 6rd, should also be
> able to take advantage of A+P. For this, IPv4 in IPv4 tunneling
> mentioned by Randy does make sense.
>
> Regards,
> RD
>
>
> There are other approaches
> Le 3 oct. 09 à 01:07, Randy Bush a écrit :
>
>>> The IESG has approved the aplusp BOF for IETF 76.  The BOF will be
>>> used to discuss A+P addressing and forwarding as it applies to
>>> dual-stack lite [draft-ietf-softwire-dual-stack-lite-01].
>>
>> excuse?  folk such as the wireless gang have a very different need
>> for
>> it, and an architecturally different spin.  some are using ipv4 over
>> ipv4 tunneling.  some are using very different signaling mechanisms.
>> etc.
>>
>> currently, ds-lite has been restricted to one view of provisioning.
>> this is a pretty restricted view.
>>
>> randy
>> _______________________________________________
>> shara mailing list
>> [e-mail]
>> https://www.ietf.org/mailman/listinfo/shara
>

_______________________________________________
shara mailing list
[e-mail]
https://www.ietf.org/mailman/listinfo/shara
_______________________________________________
shara mailing list
[e-mail]
https://www.ietf.org/mailman/listinfo/shara