Integrity protection of fetches

[Matt McCutchen]

On Fri, 2010-08-06 at 11:29 -0500, Steve Bonneville wrote:
> i.grok at comcast.net wrote:
> > Ideally (from this perspective), the host would validate the response itself.
> 
> Exactly, if sshd is sufficiently paranoid it should make a query with
> CD set in the request and do all the validation client-side.  If you let 
> your nameserver do the validation, I think it's still possible to MITM 
> this by messing with the communication between the stub resolver and the 
> name server, which isn't secured.

Not to mention that one has to trust one's own nameserver, which is a
bad idea when using a public wireless access point.  In order to achieve
the same security properties as SSL/TLS as commonly practiced, local
validation is the way to go.

-- 
Matt