At Mon, 18 Jan 2016 11:28:56 +0000,
Sara Dickinson <[email protected]> wrote:

> > Some initial comments on the 00 version follow:
> >
> > - Section 4.2
> >
> >      +-----------------------+------------------+-----------------+
> >      |     Usage Profile     | Passive Attacker | Active Attacker |
> >      +-----------------------+------------------+-----------------+
> >      |       No Privacy      |        N         |        N        |
> >      | Opportunistic Privacy |        P         |      N (D)      |
> >      |     Strict Privacy    |        P         |        P        |
> >      +-----------------------+------------------+-----------------+
> >
> > How can we detect an active attack in Opportunistic Privacy?  For
> > example, if an active attacker pretends to be the recursive server
> > that just doesn't support privacy, can't we distinguish it from a
> > genuine server that actually doesn't support it?  Perhaps it's
> > described in the following part of Section 5:
> >
> >   [...]  This is useful for
> >   detecting (but not preventing) active attack, and for debugging or
> >   diagnostic purposes if there are means to report the result of the
> >   authentication attempt.
> >
> > but it's still not very clear to me.
>
> Yes you are correct this section touches on the issue. The scenario where 
> detection is possible is if a client is configured with either a domain name 
> or SPKI pinset for a given server then the expectation is that the server 
> _is_ a DNS privacy enabling server. So if the authentication fails in Strict 
> Privacy the connection is aborted. However in Opportunistic Privacy the 
> client can choose to connect anyway, but it is aware that the server does not 
> authenticate and so it should consider an active attack in progress (worst 
> case - of course it could be a config error…).

Maybe it's just a matter of wording, but this doesn't really sound
like "Detection" to me, in that the client can't (100%) know if it's
an attack or not (that's the same for Strict Privacy, but we can
surely say it provides "Protection" in that case due to its hard-fail
nature).  But...

> We plan to update the discussion of Opportunistic Privacy in the next version 
> - thanks for this comment.

...since this may be a wording issue, I'll see how the revised version
reads.

--
JINMEI, Tatuya

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

Reply via email to