A comment of Olafur's has triggered me to write something I like about the 
charter, and also something in support of Stephane.  Olafur wrote;

> So I think the charter is right in saying “will focus on last mile” and check 
> if that solution will scale to other cases. 


The charter uses the noun “mechanisms” not “solutions, and doesn't to indicate 
the development of single one-size-fits all solution, as I read it.  It also 
makes explicit in the milestones that multiple “solutions” might be developed.  

Stephane’s existing draft about the problem statement has done a great job in 
leading us to understand that there are varied operational realizations that 
need to be served by IETF’s work here.

Operationally end-systems reach the iterative resolver and beyond in different 
ways.  Taking just two, there’s the case in which a stub and iterative resolver 
are both running on the same computer, and the case in which many end-systems 
reach the iterative resolver through enterprise name system management of some 
kind.  In both cases, you can see that the end-systems are subject to having 
their queries linkable (in a privacy-revelation sense) and subject to 
compromise of their DNS private exchange, even if some mechanism for 
confidentiality of the stub-to-iterator is present.  

I’d like to see the working group propose and specify whatever the needed 
deployable mechanisms are to provide the end-system(s) with DNS private 
exchange, and not start with a mechanistic boundary.

Best regards,

Allison



On Oct 6, 2014, at 11:26 AM, Olafur Gudmundsson <[email protected]> wrote:

> 
> On Oct 6, 2014, at 8:44 AM, Stephane Bortzmeyer <[email protected]> wrote:
> 
>> [Keep [email protected] in the loop only if it is substantive comments on
>> the WG creation, please]
>> 
>> On Fri, Oct 03, 2014 at 10:38:35AM -0700,
>> The IESG <[email protected]> wrote 
>> a message of 68 lines which said:
>> 
>>> The primary focus of this Working Group is to develop mechanisms
>>> that provide confidentiality between DNS Clients and Iterative
>>> Resolvers,
>> 
>> I do not see why the group is limited to this point. 1) Some technques
>> (such as hop-to-hop encryption) work exactly the same for this case
>> and the case of resolvers<->authoritative. 2) The problem of data
>> gathering by authoritative name servers is as serious as the problem
>> of sniffing by third parties between a stub client and a resolver, and
>> should be addressed at the same level.
>> 
>> 
> 
> Well different techniques might be “better” in the two cases, i.e. connection 
> from client to Recursive resolver 
> may only be kept open for a short time while the connection from Recursive 
> Resolver to a BIG DNS data provider 
> might be always-on. 
> So I think the charter is right in saying “will focus on last mile” and check 
> if that solution will scale to other cases. 
> 
>       Olafur

_______________________________________________
dns-privacy mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/dns-privacy

Reply via email to