[openstack-dev] [Nova][FFE] requesting FFE for LVM ephemeral storage encryption

Sean Dague sean at dague.net
Thu Sep 4 17:20:07 UTC 2014


On 09/04/2014 11:48 AM, Daniel P. Berrange wrote:
> On Thu, Sep 04, 2014 at 11:41:48AM -0400, Dan Genin wrote:
>> I would like to request a feature freeze exception for
>>
>> 	LVM ephemeral storage encryption[1].
>>
>> The spec[2] for which was approved early in the Juno release cycle.
>>
>> This feature provides security for data at-rest on compute nodes. The
>> proposed feature protects user data from disclosure due to disk block reuse
>> and improper storage media disposal among other threats and also eliminates
>> the need to sanitize LVM volumes.  The feature is crucial to data security
>> in OpenStack as explained in the OpenStack Security Guide[3] and benefits
>> cloud users and operators regardless of their industry and scale.
>>
>> The feature was first submitted for review on August 6, 2013 and two of the
>> three patches implementing this feature were merged in Icehouse[4,5]. The
>> remaining patch has had approval from a core reviewer for most of the Icehouse
>> and Juno development cycles. The code is well vetted and ready to be merged.
>>
>> The main concern about accepting this feature pertains to key management.
>> In particular, it uses Barbican to avoid storing keys on the compute host,
>> and Barbican at present has no gate testing.  However, the risk of
>> regression in case of failure to integrate Barbican is minimal because the
>> feature interacts with the key manager through an*existing*  abstract keymgr
>> interface, i.e., has no*explicit*  dependence on Barbican. Moreover, the
>> feature provides some measure of security even with the existing
>> place-holder key manager, for example, against disk block reuse attack.
>>
>> For all of the above reasons I request a feature freeze exception for
>> LVM ephemeral storage encryption.
> 
> I'm happy to sponsor this, since I've positively reviewed it and it is
> a pretty well isolated feature so risk to other existing code is low.

I'm happy to sponsor this one. The last patch is pretty straight
forward, I think one change around the way barbican is included should
make it ready (reviewed yesterday).

	-Sean

-- 
Sean Dague
http://dague.net



More information about the OpenStack-dev mailing list