Re: [DNSOP] [Ext] Re: New draft for helping browsers use the DoH server associated with a resolver

[Vladimír Čunát <[e-mail]>]

On 08/24/2018 04:56 PM, Paul Hoffman wrote:
>> Figuring out what SNI to use using insecure DNS sort of negates any advantage
>> TLS authentication offers.
> I don't understand this at all. The SNI is the host in the URI template.

There remains the general problem with bootstrapping a "secure" channel
without using another secure anchor.  DHCP has the same problem - it's
not validable (reportedly, I don't know the protocol well).

You get a name (for SNI) and thus you can verify that you connect to
that name (https/tls), but you can't distinguish whether you got a
spoofed name in the first place, and I can't see a possible workaround
for that.  Both DHCP and this draft are essentially like not validating
identity of the other side but otherwise using a "secure" transport. 
Well, I think many people consider such opportunistic security schemes
to be noticeably better than cleartext.

Still, personally I'd probably prefer to choose someone from a list of
providers, as we might have quite a lot soon, and "I" might trust some
of the names already, and the handshake will then verify that the name
matches.  I assume this draft is meant as yet another fallback when such
user selection isn't suitable.  (Well, I'd most prefer to run a per-SOHO
or per-device validating resolver, in future maybe with TLS-forwarding
somewhere else, but that doesn't belong to this thread, and it isn't
suitable for most people anyway.)

--Vladimir