Content-Type: text/plain; charset=iso-8859-1
Mime-Version: 1.0 (Mac OS X Mail 6.6 \(1510\))
From: Russ Mundy <mundy@tislabs.com>
In-Reply-To: <52A050E7.8010405@uni-due.de>
Date: Thu, 5 Dec 2013 10:35:30 -0500
Content-Transfer-Encoding: quoted-printable
Message-Id: <C94CFC5A-3A5E-427E-B269-2457A696E2DC@tislabs.com>
References: <C0D19C51-6EA6-4EAF-B9CB-D80F673262E5@icsi.berkeley.edu>
 <52A050E7.8010405@uni-due.de>
To: perpass@ietf.org
Cc: Russ Mundy <mundy@tislabs.com>,
 =?iso-8859-1?Q?Matth=E4us_Wander?= <matthaeus.wander@uni-due.de>
Subject: Re: [perpass] A reminder, the Network is the Enemy...
Precedence: list

On Dec 5, 2013, at 5:09 AM, Matth=E4us Wander =
<matthaeus.wander@uni-due.de> wrote:

> * Nicholas Weaver [2013-12-02 17:56]:
>> Actually spoofing DNSSEC replies even with knowledge of the root key =
is going to be difficult...
>=20
> If we assume the attacker can get the private root KSK from an =
US-based
> corp, then we should also assume they can get the private root ZSK =
from
> another US-based corp. As the owner of the root ZSK also owns the keys
> for .com, the attack becomes much easier.

If we (as the IETF) make an assumption that the DNSSEC private key(s) =
are "available" to some "unauthorized entity" (govt or otherwise) =
because a significant part of a particular operation is located in a =
particular geographic region then we need to also make a similar =
assumption for any/all Certification Authorities' root private key(s) =
since the underlying cryptographic technology widely used by TLS is =
basically the same.  The DigiNotar attack, though not geographically =
related, clearly illustrates that very bad things can happen when an =
"unauthorized entity" is able to have access to and use of root private =
keys for a CA.

I've seen some references on this list saying (essentially) that it is a =
valid assumption that an "attacker" ("unauthorized entity" might be a =
better term) can get or already has the DNS root (& maybe .com) private =
key.  Although I do not believe that this is a valid assumption, I do =
assert that if we (as the IETF) decide to make such an assumption =
relative to DNS/DNSSEC then we must make the same assumptions about =
"unauthorized entities" being able to access private root key(s) for =
any/all CAs.  I'm not sure how the IETF would somehow factor =
geopolitical boundaries into defining protocol assumptions, I suspect =
that any useful results would probably take longer than it's taken to =
design, redesign, redesign and begin deployment of DNSSEC :-).

OTOH, if there is real interest and need to change and/or enhance the =
security operations &/or protocols for the DNS or CA realms, having =
concrete proposals (such as  =
draft-grothoff-iesg-special-use-p2p-names-00.txt) is much more useful =
than trying to reach agreement on assumptions like the above (& other =
earlier email assertions).


Russ

>=20
> Regards,
> Matt
>=20
> --=20
> Universit=E4t Duisburg-Essen
> Verteilte Systeme
> Bismarckstr. 90 / BC 316
> 47057 Duisburg
>=20
> _______________________________________________
> perpass mailing list
> perpass@ietf.org
> https://www.ietf.org/mailman/listinfo/perpass