Whew! January was a busy month. In addition to my usual CMSWire columns (my first of the year was about the BDL role in open source), I spent time talking with journalists, working on a new research paper on Service Mesh for container clusters, and finished a paper on Cloud Foundry vs. Kubernetes. Busy busy busy.
At the beginning of the new year, I was quoted in a blog entitled “20 Developers and Kubernetes Experts Reveal the Biggest Mistakes People Make During the Transition to Kubernetes” It’s nice to be called a Kubernetes expert but I wouldn’t call myself that. Kelsey Hightower is an expert; I’m an observer. Still, I stand by my quote about one of the big mistakes when adopting Kubernetes which was “From our vantage as outside observers, it’s trying to apply Kubernetes to all applications.”
An article that I was previously quoted in came out in French. Originally published in mid-December as “Knative project stokes interest in event-driven IT ops” it came out in January in the French language version. I took French in high school and can still read it enough to decipher a wine bottle (shows where my priorities are) but do not speak it. I assume that “Knative : les entreprises montrent un début d’intérêt” quotes me correctly.
More talk about open source later in the month. More accurately, open core. Open core refers to companies that open source their core technology but maintain control over the project while adding “enterprise” features to the product they sell. “Uncertain future of open core software puts companies at risk” talks about the problems these companies have and the advantages of vendor supported open source.
Expect more of me in the press in the coming months.
I also completed a new research paper which compares Cloud Foundry and Kubernetes as the basis of cloud native platforms. I dispel the myth that it must be one or the other. I expect that to be released within the next month.
And keep your eye out for a major research paper on service mesh technology. A component of microservices architectures, a service mesh is critical to enterprise container clusters and other microservices implementations. Look for it in April just ahead of Cloud Foundry Summit in Philadelphia.
And you wonder why I haven’t been blogging here much.
Containers and Kubernetes Become Enterprise Ready
In case there was any doubt about the direction containers and Kubernetes are going, KubeCon+CloudNativeCon 2018 in Seattle should have dispelled them. The path is clear – technology is maturing and keeps adding more features that make it conducive to mission critical, enterprise applications. From the very first day the talk was about service meshes and network functions, logging and traceability, and storage and serverless compute. These are couplets that define the next generation of management, visibility, and core capabilities of a modern distributed application. On top of that is emerging security projects such as SPIFFE & SPIRE, TUF, Falco, and Notary. Management, visibility, growth in core functionality, and security. All of these are critical to making container platforms enterprise ready.
If the scope of KubeCon+CloudNativeCon and the Cloud Native Computing Foundation (CNCF) is any indication, the ecosystem is also growing. This year there were 8000 people at the conference – a sellout. The CNCF has grown to 300+ vendor members there are 46,000 contributors to its projects. That’s a lot of growth compared to just a few years ago. This many people don’t flock to sinking projects.
Despite all the growth in ecosystem and capabilities, there were still a fair number of container-curious people who were at KubeCon+CloudNativeCon. Their companies sent them to KubeCon+CloudNativeCon because they just beginning to explore containers and Kubernetes. They had a lot of questions about the viability of containers and microservices in their more demanding environments, especially regulated environments. Many of the questions I was asked, especially about visibility within clusters and security, were important discussion points. Some of the doubts were a smokescreen for organizations that resist change. It was obvious that they were looking for an excuse to stick with old ideas.
Another issue holding back container architectures is confusion in the market. CloudFoundry, serverless platforms, and Kubernetes platforms overlap and use similar technology, namely containers. Since vendors will often present these as competing platforms, depending on what they sell, it presents the market as more fragmented then it is. Even within technologies there is a lot of confusion. Take serverless computing. Ask ten people what serverless is and you will get eleven different responses. Some vendors want to make it a marketing label they can slap onto anything to make it shiny and new. This makes life very confusing for an enterprise IT professional trying to design next generation applications.
Some of this confusion is just an artifact of a lifecycle problem. Five years ago, there were several competing container formats from Docker, Rancher, CoreOs and others. That has changed. Containers have coalesced around a common image format. Container engine vendors are no longer competing on the basics but on performance and security layered over standard runtimes such as containerd.
No one is advocating change for the sake of change. We are at a point, however, where the demands of modern applications require a new architecture. Kubernetes represents an excellent platform for highly distributed applications where portability, performance, and development lifecycle problems are easily managed. The future of containers and Kubernetes as the base of the new stack was on display at KubeCon+CloudNativeCon and it’s a bright one. Expect to see more enterprise applications that rely on rigorous architectures to be Kubernetes.