[panarky]

How can Docker claim that Rocket does not welcome the notion of portability? Portability and composability are at the center of the Rocket announcement.

From the Rocket announcement:

  simple and open specifications for a portable container format...

  fill the gap for companies that just want a way to securely and
  portably run a container

From the Docker response:

  these vendors want to create orchestration solution (sic) that are
  tailored for their particular infrastructure or offerings, and do
  not welcome the notion of portability

Is this just FUD and sour grapes?

[SEJeff]

I think the rocket vs docker idea of open specifications are very different. Docker is fundamentally an API. In theory, you could create a C# version of docker that implements the docker api ontop of windows (when Microsoft adds namespacing support) and it would "just work TM". The same could be said for FreeBSD or anything else really.

Rocket looks very cool, and while it certainly does prevent some of the issues with docker security, it misses some of the benefits... like a remote API as there isn't a rktd, and many people will undoubtedly write them. So in that regard, I'd have to say rocket offers less openness.

With the current docker design, I see no reason why libcontainer couldn't add a rocket or lxd backend and be done with it however. As a sysadmin/developer to builds things, I see this as healthy for the container ecosystem even if it does certainly come off as a bit uncool from @phillips and the coreos crew. I'm kind of biased in that I think the only good multi-host docker / container manager will ultimately be mesos with kubernetes ontop.

[mrisse]

Docker isn't claiming that Rocket doesn't welcome the notion of portability. They are saying that it would be a shame if portability was lost in the world of multiple container, distributed applications.