To celebrate the 1.0 release of its container-orchestration system Kubernetes, Google's engineers did something that falls in line with recent moves in the container world: They gave it away.
But Google may still benefit from Kubernetes with an offering akin to the open-standard public cloud that's been dreamed of for some time.
How the pieces fall into place
With Kubernetes now under the umbrella of the newly formed Cloud Native Computing Foundation, Google clearly intends for everyone to benefit from its container-managing wünderkind.
Kubernetes is mostly associated with managing containers across clusters of machines -- launching, grouping, allocating resources, balancing loads, and segmenting containers by usage (such as dev vs. ops). Third-party applications like Jenkins or Dynatrace's Docker plug-in can now depend on Kubernetes's API to be stable. But many of the new features, such as debugging and inspection tools, and support for stateful applications by way of a variety of volume types, involve apps themselves.
Because the entire stack, containers and management tools alike, is open source, Google's larger ambitions may revolve around allowing the newly container-centric to relocate their workloads more easily into its cloud without feeling like lock-in is taking hold.
Crucially, this doesn't include the applications alone, but the management stack as well -- the secret sauce that allows an app to run well, scale efficiently, recover from disaster without blinking, and be genuinely useful in the long run. With open source management components, it's universally adoptable, and with far less clunkiness and overhead than projects like OpenStack.
Open to all
Google's competition isn't waiting around to offer similar features. Amazon and Microsoft have both made massive moves toward containers, the latter taking the unprecedented step of adding low-level functionality to Windows Server and Azure.
Google enjoys one advantage: It's already been running the very features it's now sharing with the world and for which it's soliciting feedback. Microsoft's cloud plan has something of the same smell: Offer a substrate for cloud computing (Azure Service Fabric) that's the same locally and remotely. But Microsoft's plan is tied heavily to the Microsoft stack and to existing Microsoft customers, where it'll have the most appeal. Kubernetes and containers bring no such baggage.
In the long run, Google loses nothing by having Kubernetes adopted by everyone and anyone. The mere fact that someone offers it in a cloud service won't be a standout feature; it'll be how well it can be scaled in that service, what kind of conveniences can be provided around it, and how appealing the package is to devs and enterprises.
If Google's value-add with developer-centric offerings (such as Google Cloud Repositories) helps tip the balance in its favor, it may find itself in the enviable position of being the first to offer an open-standard hybrid cloud that's a real product, not simply marketing or good intentions.