In article <[email protected]> you write:
>Please see
>   
> https://tools.ietf.org/html/draft-pp-recursive-authoritative-opportunistic-01
>This isn't a WG document yet, but if the WG wants it, I think it could work 
>well within the charter, and with the discussion of 
>draft-ietf-dprive-phase2-requirements.

I'm having some trouble following this draft since it looks like parts
of the text got scrambled, e.g.:

   In the opportunistic encryption described here, there is no need for
   the recursive resolver to authenticate the authoritative server
   because any authentication failure does not cause the TLS session
   from being set up.  

I get the gist of it, caches consult an oracle to see whether to use
DoT, and you have to find your own oracle. That seems uncontroversial,
but I fear it is so uncontroversial that it won't be useful.

As far as whether to validate the server's certificate, that seems to
me to be a policy decision based on your threat model. If your threat
is traffic being snooped in transit, you don't, if it's traffic being
stolen by hostile servers, you do. 

We leave for later the question of to what extent authenticating the
authoritative server might be a substitute for DNSSEC (keeping in mind
we know where to find dnscurve if that's what you want.)

R's,
John

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

Reply via email to