Re: [v6ops] Fwd: 464XLAT Trial Deployment Report

[Behcet Sarikaya <[e-mail]>]

On Fri, Feb 3, 2012 at 4:54 PM, Cameron Byrne <[e-mail]> wrote:
> On Fri, Feb 3, 2012 at 2:37 PM, Behcet Sarikaya <[e-mail]> wrote:
>> Hi Fred,
>>
>> I have technical concerns:
>>
>> - RFC 6146 does use DNS64 as an integral component. In this draft
>> DNS64 is not used however RFC 6146 is referenced. I think that this
>> point should be clarified instead of just giving a reference to 6146.
>>
>
> RFC6146 is stateful NAT64.  It does not require DNS64.  They are
> completely decoupled.  RFC6146 can function fine without DNS64.
> draft-mawatari-v6ops-464xlat can also function fine without DNS64, in
> my particular implementation i choose to use DNS64 since it reduces a
> layer of translation in most of *my* cases.


My point was to ask you to add at least one paragraph explaining how
DNS64 is avoided.

>
>> - As you will remember, in 3GPP-IETF workshop in 2010, pnat proposal
>> was discussed in depth. I think 464XLAT is very much like pnat. The
>> workshop did not endorse such approaches because of UE modification
>> requirement. Instead, dual stack network was endorsed.
>>
>
> PNAT, AFAIK, had an on-host DNS64 ENR function and specified ALGs.  It
> was overall a bit more complicated, and eventually that work evolved
> into the BIH work that has been approved as an RFC.   PNAT was also
> standards track work, as where this draft is informational.  From my
> perspective, the BIH work does not solve my stated problem scope since
> it is only scoped for IPv4 applications communicating with IPv6-only
> servers.
>

I know the pnat story.


>> - Since Release 8, 3GPP has adopted v4v6 or dual stack PDP Context
>> which is not mentioned in the draft.
>>
>
> Why would there be a mention of that?  The scope here is IPv6-only
> access networks, not dual-stack networks.  That said, there are very
> few Release 8+ networks deployed globally. And, dual-stack faces the
> challenge of provide IPv4 addresses that are no available.
>

Because in Section 3 you have this:

This is especially cost-effective in wireless 3GPP
       network that would otherwise require two separate PDP connections
       to support IPv4 and IPv6.



Regards,

Behcet