Red Hat has officially announced support for the Open Container Project, an effort that will be guided by the Linux Foundation with the objective of standardizing two key ways in which containers interact with their host operating systems: the container format and the runtime.
"We have been investing in containers for many, many years," said Lars Herrmann, general manager of Red Hat Enterprise Linux and container strategy. "But we have seen conflicting ideas on concepts as to how the architecture should look and how technologies should interact, and with this we have seen some early signs of fragmentation." It is no surprise that an organization so deeply rooted in the concepts of open source technology would see a certain degree of standardization as a solution to the problem of container fragmentation.
The enemy of innovation
Currently, the application container segment is the most exciting in the enterprise computing community, and the reason is the staggering number of advancements and innovations that have happened there. But nothing puts the brakes on innovation faster than a lowest common denominator standard that forces each and every vendor to put their round pegs into square holes. So when Red Hat, the second largest contributor to the Docker platform, announces a new container standard based primarily on code donated by Docker itself, an eyebrow is raised and healthy cynicism comes to the fore.
But Herrmann insists that such skepticism is misplaced. "First of all, the most important aspect that often gets forgotten is that the container format is just a means for how a container gets moved from a to b; how it gets deployed on a system and how it gets instantiated. It doesn't speak at all to what is inside the container," Herrmann said. "The huge ecosystem of value that we will see begins with what is contained within these containers, which is applications and services."
The huge ecosystem of value that we will see begins with what is contained within these containers, which is applications and services.
Lars Herrmann, general manager of Red Hat Enterprise Linux and container strategy
So what will be the key areas of differentiation between application container vendors in a new age of standardization? Orchestration is a big one. How are containers running on a single host clustered together into a global network? That's not a problem solved by standardizing the container format, and the big players -- be it the Kubernetes approach or the one used by Docker's Swarm project -- standardization in that arena isn't on anybody's long-range radar. "Orchestration and management are areas that we will continue to see innovation given the different usage scenarios, use cases and audiences," Herrmann said. "But now all of those different solutions can rely on a common foundation that is defined by the standards set forth in the Open Container Project."
Opportunities for differentiation in the container segment
Of course, orchestration is largely a technical issue dealing with operations that take place at a low level on the application stack. But as an enterprise, operators and administrators aren't simply interested in a system that works; they need one they can work with, and that means having access to a variety of management tools that make interacting with, automating and understanding what's going on with the low- level architecture. "As an enterprise, you need the management tools that give you a UI or an API against which you can program and automate," Herrmann said.
Herrmann suggests that significant differentiation will occur on the delivery of what he describes as specialized services. "Containerized infrastructure is inherently distributed and service oriented, so there is lots of room for specialized solutions helping with things like logging or monitoring; things that are in the traditional management domain, but things that are not necessarily concerned with aggregating containers." This means vendors have the time to create best-of-breed solutions that address a given problem or use case, all while having confidence in the mechanism used to interact with the underlying platform and know exactly what is going on with the host system.