[openstack-dev] Move keypair management out of Nova and into Keystone?

Bryan D. Payne bdpayne at acm.org
Tue Jul 2 19:47:08 UTC 2013


> If you do not trust keystone to give you the right information you have
> already lost as keystone is used (afaik) to check for authorization
> anyway.
>

This is true.


> Can you be a little bit more explicit on the threat model you have in
> mind and what guarantees Barbican would give you that would make it more
> suitable to store public key than Keystone ?


I'm concerned about malicious tampering with the keys.  If the keys are
then use for validating that a user is presenting the correct private key,
this could result in an instance compromise.  Yes, if someone tampers with
other data in keystone then it could result in a compromise as well.  This
is true.

As I think about this some more, I think the best way to frame it is that
-- for me -- key data and user / password data are two different classes
that may have different security requirements.  It is nice to not mix the
two, IMHO.  However, I can appreciate the simplicity that comes with just
not using Barbican and throwing everything in Keystone.

WIth this in mind, I do like the idea of having Keystone return a pointer
to the key location as URL.  This can be a ref back to a Keystone route.
 Or it can be a ref to a Barbican route.  This would be most flexible and
allow people to fullfil different security and auditing requirements.

-bryan
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack-dev/attachments/20130702/e24c33eb/attachment.html>


More information about the OpenStack-dev mailing list