[clug] Virtual Machine security, esp. under 'Live Migration' and Disaster Failover

[Daniel Pittman]

steve jenkin <sjenkin at canb.auug.org.au> writes:

> Stumbled on this VMware specific product using VMware's 'VMsafe' -
> hypervisor kernel modules for security.
>
> In a VM farm, how do you do the same security 'stuff' you do with
> Physical world? Especially under Live Migration or Failovers..
>
> Any solutions in XEN or OSS world?

In part, I take the solution that Altor Networks describe as "bad", which is
to treat the individual machines in our VM system[1] as if they were actual
physical hosts.  In most cases this gives all the benefits they claim their
in-hypervisor firewall does.

I also use firewalling on the host Linux instance, which is effectively the
in-hypervisor stuff that they are listing.


We don't currently use the capability to inspect processes and files within
the OpenVZ containers from the host node, but I do have a vague plan to figure
out what (if any) actual use that could be to us at a later date.

Certainly, if we were in an environment where something like tripwire was
desired using the ability to run it outside the container, and so insulated
from the software inside, would be nice.


OTOH, a container system potentially has lower security than the VMWare or Xen
hypervisor model since it only has a single kernel, so the benefits are
certainly reduced...

        Daniel

Footnotes: 
[1]  Typically, OpenVZ containers, but a few VMWare Server, soon to be KVM,
     machines as well.

-- 
✣ Daniel Pittman            ✉ daniel at rimspace.net            ☎ +61 401 155 707
               ♽ made with 100 percent post-consumer electrons