Re: [v6ops] draft-byrne-v6ops-clatip-00

[Lorenzo Colitti <[e-mail]>]

On Fri, Oct 11, 2013 at 10:32 AM, cb.list6 <[e-mail]> wrote:

> I wrote a draft to expand the scope of the ds-lite v4 range to include
> 464xlat in its scope
>
> Please have a look and let me know your thought on adopting and moving
> this forward.
>
> http://tools.ietf.org/html/draft-byrne-v6ops-clatip-00
>
Full disclosure: I
implemented<https://android.googlesource.com/platform/external/android-clat/+/d6ef91bef28020ec019a64561e8fb5e353e42980>
this
in Android, which uses 192.0.0.4, so I obviously support it. :-)

The reason I think this makes sense is because the semantics of the 464xlat
IPv4 address are very similar to the semantics of the B4 IP address. Like
DS-Lite, the 464xlat IPv4 address never appears on the wire, and it clearly
identifies transitional IPv4 connectivity. Again like DS-Lite (but unlike
other transition mechanisms such as MAP) the 464xlat IPv4 address is
globally identical for all devices that implement the techology.

Using an IPv4 address in the DS-Lite range allows other standards and code
which need to deal with the properties of IPv4 addresses (for example,
address sorting as documented in RFC 6724) to only support one
"transitional IPv4" range instead of separately implementing mostly
identical behaviour for DS-Lite and 464xlat (and possibly other similar
IPv4-in-IPv6 transition mechanisms that have not yet been defined).

If this document goes forward it might make sense not to have it refer to
464xlat explicitly but to specify under what conditions a transition
mechanism can use this range. For example, instead of explicitly referring
to 464xlat, the document might say that this range can be used for any
transition mechanism which uses IPv6 to provide IPv4 and where the IPv4
address is the same for all hosts and never appears in the wire.