OpenStack Havana is out and includes native Docker support
blog.docker.io
blog.docker.io
That is, say I start with an Ubuntu image. Then I run an application on it on a Docker cloud. Then I install Python 3.0 and NumPy on it and take a snapshot. And then I want to run the new app with these libraries. I think when you deploy, or when images are being sync'd to machines, they can use the differential compression based on AUFS snapshots. So instead of transferring 300 MB, maybe you will only transfer 30 MB.
That would seem logical based on what I've read. But I have never actually gotten Docker to work, so I don't know. Would appreciate any comments on this.
Also, the Docker workflow is appealing to (at least) some people; this helps them to integrate Docker and OpenStack together.