On Tue, Jun 16, 2015 at 10:34 AM, Sam Hartman
<[email protected]> wrote:
>
> So, NVO3 is supposed to provide isolation for a TS.
> TSes aren't really trusted as I understand it.
>
> However, we'd like to provide the ability to secure some parts of the
> system against compromises that affect more than just a TS.  In
> particular, we'd like to prevent a compromise of one node in the
> virutalization environment from giving you a network-wide compromise.
> You will doubtless be able to affect everything going through/hosted on
> that node, but this shouldn't imply a compromise of the mapping layers
> or data center.

Understood, but the document says (Section 4):
".. attacks are classifed ..  into three
   categories: the attacks from compromised NVO3 devices (inside
   attacks), the attacks from compromised tenant systems, and the
   attacks from underlying networks (outside attacks)."

Seems like a compromised TS is also an inside attacker (and
vice-versa?) so the protection measures for both of these should
always be the same? If they are not always the same, it might help
to have the doc explain when they would be different, and what
differences to take into account, when coming up with the security
protections.

--Sowmini

_______________________________________________
nvo3 mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/nvo3

Reply via email to