The use of software containers like Docker has been one of the biggest industry trends this past decade. Hitherto, the focus has been to educate the development community on the capabilities of container technology and how enterprises might use container orchestration tools effectively. But it would appear that an event horizon in terms of container awareness has been breached, as familiarity has increased to the point where organizations have moved beyond simply entertaining the thought of bringing containers into the data center and have moved forward into actual adoption. But before companies actually deploy software into containers, the question remains about which software container, and complimentary set of container orchestration tools, an organization should adopt. Options such as Swarm, Cloud Foundry and Mesos make strong arguments for themselves so the container technology landscape is a difficult one to navigate.
Container orchestration with Kubernetes
Of course, Craig McLuckie, CEO of Heptio, is confident in the horse his organization is betting on to create cloud-native applications. “Kubernetes over the last three and a half years has emerged as the standard.” Mark Cavage, VP of Product Development at Oracle, also gives Kubernetes the thumbs up. “It’s the right open source option that abstracts away clouds and infrastructure, giving you a distributed substrate to help you build resilient microservices.”
Kubernetes has been compared favorably with other container orchestration solutions before, with one of its big selling points the vendor-agnostic nature of its open source availability. McLuckie poins out the lack of vendor lock-in as an appealing benefit. Other experts have praised Kubernetes for working well with Docker while requiring only a one-tier architecture. Because of its flexibility, developers who use Kubernetes may be less likely to face the situation of writing applications to match the expectations of a vendor-driven container orchestration platform rather than writing applications in the way that makes the most sense for what their organization hopes to accomplish.
McLuckie speaks highly of Kubernetes’ ability to simplify ops with containers that are a hermetically sealed, highly predictable and portable unit of deployment. But deployment is just where the real work begins. Keeping the containers running and automating a container based architecture without a hiccup is where Kubernetes shines. “By using a dynamic orchestrator, you can rely on an intelligent system to deal with all of those operational mechanics like scaling, monitoring, and reasoning about your application.”
Increased uptime and resiliency are obvious benefits of having a well-orchestrated container system. And since the self-contained units run the same way on a laptop as they do in the cloud, they can make development and testing easier for transitioning DevOps teams. End users also enjoy the perks of Kubernetes without realizing it, since the orchestration tooling can perform automatic rollout and rollback of services live without impacting traffic, even in the event of a failure of the new version.
Container orchestration and middleware
One interesting point McLuckie makes pertains to how middleware is being “teased apart” as a result of containerization. “Systems like Kubernetes are emerging as a standard way to run a lot of the underlying distributed system components, effectively stitching together a large number of machines that can operate as a single, logical, ideal computing fabric.” As a result, “A lot of the other functionality is getting pushed up into application level libraries that provide much of the immediate integration you need to run those applications efficiently.”
What might the outcome of this fragmentation of middleware be? According to McLuckie, it could open Java apps to a polyglot world by making it simpler to peacefully coexist with other languages. In addition, it would make dependencies easier to manage, even supporting the ability to run something like Cassandra DB on the same core infrastructure and underlying orchestrator as the Java app that relies on that database.
Cutting down on system complexity could be what makes Kubernetes and containerization itself attractive to larger organizations in the long run. “This could address enterprise concerns for governance, risk management, and compliance with a single, consistent, underlying framework.” And enterprises that prefer on-premise or hybrid clouds could use this approach just as readily as those that are fully cloud-reliant.
With container orchestration tools becoming more sophisticated with time, don’t expect the hype around container technology to die down in 2018. It will only intensify, with the focus moving away from education and awareness and instead towards adoption and implementation.