Re: [Int-area] [ietf-privacy] NAT Reveal / Host Identifiers

[<[e-mail]>]

Hi Stephen,

Please see inline.

Cheers,
Med

>-----Message d'origine-----
>De : Stephen Farrell [mailto:[e-mail]]
>Envoyé : vendredi 6 juin 2014 17:59
>À : BOUCADAIR Mohamed IMT/OLN; Ted Lemon
>Cc : Brian E Carpenter; [e-mail]; [e-mail]
>Objet : Re: [Int-area] [ietf-privacy] NAT Reveal / Host Identifiers
>
>
>Hi Med,
>
>On 06/06/14 12:41, [e-mail] wrote:
>> [Med] I'm not sure about this Ted. There are other initiatives that
>> try to solve the issue for particular use cases, see for instance
>> this effort for HTTP:
>> http://tools.ietf.org/html/draft-ietf-appsawg-http-forwarded-10. The
>> rationale of the "host identifier scenarios" document is to group all
>> use cases suffering from the same problem instead of focusing on one
>> single case. This allows having a big picture view of the problem
>> space.
>
>I think XFF is actually a good example of why we ought not adopt
>this work.

[Med] I provided "Forward" as an example to illustrate there is a need to have a big picture view rather than focusing on specific use case. This draft does not suggest to adopt XFF or Forward at all. There is a need to understand the problem space and identify deployment scenarios where encountering complications.

>
>XFF is widely deployed already but somewhat flakey in terms of
>interop so the authors of the above draft aimed to produce a spec
>that just addressed interop. (*) That was adopted by an area WG
>without (IMO) ever really considering the privacy implications,
>and definitely without any effort having been made to develop a
>more privacy-friendly mechanism (which could have been done, again
>IMO) to solve what were the claimed use-cases.

[Med] Wouldn't be this effort an opportunity to promote those solutions you are advocating for? The proxy use case (not limited to HTTP) is listed as a typical scenario. If there are other better solutions that solves your privacy concerns, why not documenting them? 

 By the time it
>got to the IESG that was in practice unfixable and after some
>fairly painful discussions it ended up with 4 abstain ballots,
>including mine. [1] (For those who quite reasonably don't need
>to care about IESG balloting, an abstain means approximately
>"yuk.":-)
>
>Of course that all also pre-dated BCP188 and the last year's
>shenanigans so I'd hope we have learned to not keep doing that.
>
>I'd be delighted if those who could get a better solution
>implemented/deployed were to attempt to try to address the
>privacy issues with XFF but it seems that at least in that
>case relevant folks don't care (sufficiently;-) deeply about
>our privacy to go do that.
>
>S.
>
>[1]
>https://datatracker.ietf.org/doc/draft-ietf-appsawg-http-forwarded/ballot/
>
>(*) To be clear: I think the authors were genuinely just
>trying to fix what they saw as an interop problem.