A mere month after Docker and other companies formed the Open Container Project, which placed their software-containerization concepts under the control of the Linux Foundation, another major initiative involving containers is taking off -- and many of the same people are in the driver's seat and riding along.
The Cloud Native Computing Foundation (CNCF), the Linux Foundation's newest creation, aims to bring together many of the companies involved with "[cloud-native] applications or services that are container-packaged, dynamically scheduled and micro services-oriented," according to the Foundation's press release.
A new group for a new stack
Mesosphere, a participant in the CNCF, described the project's mission as aiming "to provide developers with the right set of tools for building, deploying, and managing next-generation, 'cloud-native' applications."
The CNCF's focus is wider compared to the OCP. Rather than deal with just container technology in the abstract, the CNCF aims to govern both containers and the stack of technologies around them -- e.g., Google's Kubernetes, now in its 1.0 release -- as they're used to build business apps in a public or private cloud environment. Kubernetes is being provided to the CNCF, in much the same way that Docker's container spec was donated to the OCP.
The CNCF is trying to bring a "neutral and collaborative forum" to the table to do this work, where the technology is defined with a "code-first approach" -- meaning that working reference implementations matter more upfront than standards on paper.
This echoes the way the current spate of container technologies have sprung into being, via an open source product that could be used immediately rather than a spec that needed to be implemented.
Move fast, but don't break too many things
The OCP's philosophy, where "rough consensus and running code" and "code leads spec, rather than vice-versa," is of the same breed. Unfortunately, this approach has also yielded its share of criticisms -- for example, that Docker isn't documented well enough or hasn't had enough thought applied to its security or networking models before being released as a product.
The CNCF will need to balance the need to be fast-moving -- something demanded of container technology as companies hustle to implement it as a business-accelerating measure -- against the need to provide something that can be built on without excessive investments of manpower or man-years.
Another potential pitfall is that it could become unclear who's supposed to take responsibility for what. Will responsibility for container monitoring, for instance, lie with the OCP -- now being rechristened the Open Container Initiative -- or with the CNCF?
It's likely those divisions, like the still-under-wraps governance model for the CNCF, will be worked out in time. But the process will need to be transparent to non-members, since they'll be the ones deciding whether to implement what the CNCF has to offer.