Re: [xri] Re: question about dns trust profile

[Peter Davis <peter.davis@neustar.biz>]

On Feb 5, 2009, at 2:56 PM, Brian Eaton wrote:

> On 2/5/09, Peter Davis <peter.davis@neustar.biz> wrote:
>> True, but no more so than an A record attack on the DNS for almost  
>> every
>> resource we have.  They real test, IFAIC, is a trust in the  
>> signature keys.
>> Anything else is liable to introduce attacks.
>
> I think the attack exists even with trust in the signature keys.

What i was really meaning here, is that the relying party to private  
key K accepts the policies under which the bearer of key K adheres to  
(one such policy being: "bearer of K will sign all documents with the  
same key, the signature indicates adherence (of the document) to (some  
articulated policies)")

> Consider this:
>
> Same key is used to sign two documents, A and B.
> Legitimate DNS entry specifies that resource X maps to document A.
> Spoofed DNS entry specifies that resource X maps to document B.

True, but if the key location is also in the DNS (as I indicated would  
be the case for delegated signing), both Document A's location and the  
Keys for the signature for Document A are both addressed in the  
namespace of resource x.

=peterd