extending at the edge

[Brian E Carpenter]

On 11/10/2012 12:48, Lorenzo Colitti wrote:
> On Thu, Oct 11, 2012 at 8:44 PM, Nick Hilliard <nick at foobar.org> wrote:
> 
>> >From a protocol engineering point of view, this makes me want to roll my
>> eyes and hit my head off the desk because it's damned stupid.  So what can
>> we do in the IETF to issue guidance to stop this sort of thing from
>> happening?  Do we have enough guideline documentation in the RFCs / ID
>> pipeline at this stage so that we can tell providers "don't do this because
>> it's stupid and broken"?
>>
> 
> If you believe that the operators will listen to something the IETF says,
> then perhaps a brief document that lists the reasons why a) /64 is useful
> and b) /128 -> NAT is a bad idea and won't work anyway, might have the best
> way forward.
> 
> I'm not aware of such a document. Perhaps that's because everybody at the
> IETF believes this and therefore does not see the need for it. Perhaps we
> do need it.

iirc there were attempts to educate the 3GPP world before they created
the current situation, but those attempts failed. Surely however, every
provider knows that the typical IPv4 /32 is shared via NAT, so they know
that "one address == one device" is already a broken model. However, I
see no harm in somebody drafting a short, to the point, document
for v6ops at ietf.org.

draft-byrne-v6ops-64share only covers the 3GPP /64 aspect.

RFC 6177 says that /128 is a Bad Thing but doesn't give detailed arguments.

   Brian