Re: [v6ops] draft-ietf-v6ops-ula-usage-recommendations - work or abandon?

[Brian E Carpenter <[e-mail]>]

Going back a bit:

On 04/11/2015 23:42, Gert Doering wrote:
> Hi,
> 
> On Wed, Nov 04, 2015 at 04:36:05AM +0000, Howard, Lee wrote:
>> 2. There is consensus that NPT66 is bad, for some of the reasons give in
>> various NAT44/CGN documents. It might be slightly less bad than NAT66,
>> since it has slightly fewer of the problems given in those documents.
> 
> I don't think there is consensus on that... otherwise we wouldn't have
> published NTP66.

There was rough consensus to publish NPTv6 as experimental. There was
clearly no consensus to publish it on the standards track.

But anyway: that is history now. IMHO we need text in the current draft
that not only states that ULA+NPTv6 has drawbacks, but actually summarises
them. And states that the IETF has not defined NAT66, and summarises the
additional drawbacks that ULA+NAT66 would bring. I believe that most of the
information about the drawbacks is already documented, in RFC 6296 itself,
RFC 4864, RFC 2993 and maybe elsewehere. There is a good set of
references in RFC 6296.

However, I don't think any of those references discusses the load balancer
drawback. It is mentioned in a different context at
http://tools.ietf.org/html/rfc7098#page-10. As it happens, that describes
a case where NPTv6 is definitely less harmful than NAPT. (When we say
NAT66, are we referring to NAT66 or NAPT66?)

Bing already indicated that the authors will add more or less what I suggest.
That's really the only change needed, IMHO. I'll save a detailed re-review
for the next draft.

   Brian