Re: [Softwires] Stateless implementation plan

Le 2012-02-08 à 20:49, Cameron Byrne a écrit :

> On Wed, Feb 8, 2012 at 1:23 AM, Rémi Després <despres.remi@laposte.net> wrote:
>> Hi Behcet,
>> 
>> Le 2012-02-08 à 09:46, Behcet Sarikaya a écrit :
>> 
>>> Hi Cameron,
>>> 4rd solution IMHO is more suitable for a fixed network. CPE in 4rd is
>>> not appropriate to be hosted in a UE.
>>> 
>>> I think your solution 464XLAT's mobile part is way better for your
>>> purposes. There you can put all your IPv4 resources on the PLAT box so
>>> that CLAT box is kept simpler.
>>> 
>>> In 4rd, CPEs have A+P and BR is kept "stateless" these are not so
>>> useful for your purposes, I think.
>> 
>> Note however that:
>> - 464XLAT doesn't support shared IPv4 addresses (while 4rd does)
> 
> Hmm... we may be getting off topic.
> 
> 464XLAT definitely shares IPv4 addresses on the PLAT -- That is RFC
> 6146 -- statefule NAT64.

Well, if a customer _uses_ 464XLAT, it does get a full IPv4 address (sec 6.1). 
I don't see why a customer that has both an IPv6 prefix (used for more and more of its traffic) and a full public IPv4 address would need needs to use a NAT64.

464XLAT has a DNS proxy in CLATs, which I find a GOOD feature.

Besides that, it ressembles to a subset of some solutions worked on in Softwire, restricted to full IPv4 addresses and to hub and spoke.
Relationship with these other solutions belongs AFAIK to Softwire. 

> Perhaps what you mean to say is that 464XLAT uses stateful sharing of
> IPv4 addresses as defined in RFC6146.  The public IPv4 address
> resources are decoupled from the IPv4 service deployment at the edge.
> This is more flexible, IPv4 efficient, and allows for geographic
> redundancy of the translation exit points.
> 
> So, to summarize.  Both 4RD and 464XLAT support address sharing. One
> is stateless and the other is stateful.

> 
>> - 4rd over 6rd can work, and therefore offer both IPv6 and shared-address IPv4 on an RFC1918 network, e.g. on a 3GPP IPv4 PDP (while 464XLAT cannot AFAIK).
>> 
> 
> 4RD over 6RD?
> 
> That is a lot of RD :)

IPv4 Residual Deployment over IPv6 Rapid Deployment, why not?
It is AFAIK a stateless alternative to PCP in DS-lite.

Regards,
RD

> 
> I believe the implementation report here shows that 464XLAT provides a
> good users experience on 3G networks
> 
> http://www.ietf.org/mail-archive/web/v6ops/current/msg11906.html
> 
> CB
> 
>> Regards,
>> RD
>> 
>>> 
>>> Regards,
>>> 
>>> Behcet
>>> 
>>> 
>>> On Tue, Feb 7, 2012 at 9:05 AM, Cameron Byrne <cb.list6@gmail.com> wrote:
>>>> Are the map and 4rd solutions deployable for existing networks that do not
>>>> have reserves  of ipv4 ?  My assumption is that these solutions target
>>>> existing networks that have meaningful growth and they need a v6 solution.
>>>> 
>>>> If yes, how? Any pointers within the reams of drafts I should look for?
>>>> 
>>>> In my brief and simple skimming, it appears to me that setting up one of
>>>> these solutions would require me to collapse my existing network to harvest
>>>> back the addresses so that they may be redeployed in map.
>>>> 
>>>> What would the deployment process be for an address exhausted network of 10
>>>> million subs with 10% annual growth be?
>>>> 
>>>> Cb
>>>> 
>>>> 
>>>> _______________________________________________
>>>> Softwires mailing list
>>>> Softwires@ietf.org
>>>> https://www.ietf.org/mailman/listinfo/softwires
>>>> 
>>> _______________________________________________
>>> Softwires mailing list
>>> Softwires@ietf.org
>>> https://www.ietf.org/mailman/listinfo/softwires