On Mar 8, 2014, at 7:03 AM, Brian E Carpenter <[email protected]> 
wrote:

> Dan,
> 
> On 08/03/2014 09:31, Dan Wing wrote:
>> On Mar 6, 2014, at 6:03 PM, Brian E Carpenter <[email protected]> 
>> wrote:
>> 
>>> a) Since this is fixing some of the damage done by NAT, it's
>>> really unfinished business for BEHAVE, which if iirc was a
>>> Transport Area WG. Just saying...
>>> 
>>> b) The word "privacy" doesn't appear in the draft. Discussing
>>> privacy aspects is clearly essential if there is any thought of
>>> advancing this work. Actually I doubt if such a host ID is ever
>>> going to be acceptable from a privacy point of view, unless the
>>> end system is at liberty to change it at random (like RFC 4941).
>> 
>> I interpret your statement to mean that address sharing is a desirable 
>> security property.  If that interpretation is correct, where does that leave 
>> IPv6?
> 
> I'm not sure where you found that in what I wrote.

The authors can probably pull or copy text from RFC6967's Section 3 and its 
Security Considerations.

> I think that address
> sharing is undesirable from every point of view; I suppose it
> might reduce the risk of layer 3 based tracking of users,
> but that's certainly not what I meant.

Address sharing does reduce the risk of layer 3 tracking.  Combined with port 
randomization between subscribers ("NAPT"), it can do a pretty helpful job if 
that is the intent.

> IPv6 with RFC 4941 (which
> appears to be deployed in the vast majority of IPv6 clients today)
> is fine from this point of view.

Not really.  But that is a different discussion; perhaps we should have that 
discussion somewhere.

> I don't see any difference in
> privacy effect between a randomly-changing shared-IPv4 Host ID and
> a randomly-changing IPv6 address.

The difference is scope:  can the traffic be identified to a single home (IPv6 
/64 address; IPv4 /32 address; or a PSTN telephone number), or can the traffic 
only be identified to a region within a city (HTTP caching proxy; carrier grade 
NAT).

>>> c) A hard-nosed argument is that since we want to sunset IPv4,
>>> it's time to stop working on ways of making NAT solutions work
>>> better. Is there anything in the use cases that can't be fixed by
>>> native IPv6?
>> 
>> Yes, attackers won't move to IPv6 if IPv4 provides them a superior way to 
>> hide their activities.  There are attackers already using IPv4 CGN to 
>> obfuscate themselves.
> 
> Attackers will move to IPv6 when their targets do so. I guess
> such attackers will find RFC 4941 useful too. What's good for
> the privacy of legitimate users is also good for the privacy
> of attackers.

-d


>   Brian
> 
>> -d
>> 
>> 
>>> (The use case in expired draft
>>> http://tools.ietf.org/html/draft-sarikaya-fmc-prefix-sharing-usecase-01
>>> is not at all convincing to me, especially when adding the privacy
>>> argument. It actually seems to describe a bug in 3GPP. But in any case,
>>> the draft appears to suggest mitigations.)
>>> 
>>> Regards
>>>  Brian
>>> 
>>> On 07/03/2014 05:28, joel jaeggli wrote:
>>>> Greetings int-area and hiaps-mailing-list folks,
>>>> 
>>>> I realize that this is midweek at the IETF, however this question is not
>>>> far from several discussions I've had this week.
>>>> 
>>>> I have been asked to consider AD sponsoring
>>>> http://tools.ietf.org/html/draft-boucadair-intarea-host-identifier-scenarios-04
>>>> 
>>>> In the process of  considering doing so I'd like to get some input with
>>>> respect to:
>>>> 
>>>> A. The appetite for pursuing some or any of this work in existing
>>>> working groups, and in particular within the INT area.
>>>> 
>>>> B. A consensus basis for moving beyond RFC 6269 into active work in this
>>>> area.
>>>> 
>>>> C. How we address concerns raised by the IETF community expressed
>>>> through  draft-farrell-perpass-attack when evaluating scenarios and
>>>> beginning to address requirements and solution-space.
>>>> 
>>>> Obviously these are complex questions and I do not expect that we will
>>>> arrive at answers easily nor does work on this or other drafts depend on
>>>> answering them, however it's part of the dialog.
>>>> 
>>>> Thanks
>>>> joel
>>>> 
>>>> 
>>>> 
>>>> ------------------------------------------------------------------------
>>>> 
>>>> _______________________________________________
>>>> Int-area mailing list
>>>> [email protected]
>>>> https://www.ietf.org/mailman/listinfo/int-area
>>> _______________________________________________
>>> Int-area mailing list
>>> [email protected]
>>> https://www.ietf.org/mailman/listinfo/int-area
>> 
>> .
>> 

_______________________________________________
Int-area mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/int-area

Reply via email to