As more IT organizations become familiar with Kubernetes clusters as a mechanism for orchestrating containers, they are now starting to apply that concept to other types of software artifacts.
IT teams are already taking advantage of Kubernetes controllers, a mechanism to manage state, and Operators, a method of packing applications, to extend the Kubernetes control plane to manage everything from databases to virtual machines. There is also already an effort to extend the reach of Kubernetes to artifacts built using the WebAssembly (Wasm) format that is being spearheaded by Cosmonic.
Upbound has even gone so far as to develop Crossplane, an open source project being advanced under the auspices of the Cloud Native Computing Foundation (CNCF) that provides a control plane based on Kubernetes application programming interfaces (APIs) that can be applied to any platform.
In effect, Kubernetes is starting to evolve into vendor-neutral platform for managing a wide range of components that has major implications for how IT environments will ultimately be managed.
At the heart of the Kubernetes control plane are controllers that continuously monitor resources to ensure a system’s current state automatically aligns with the defined desired state. For example, a Kubernetes controller ensures the specified number of replicas of ab application is always maintained. If a pod crashes or becomes unresponsive, a Kubernetes deployment controller will recognize the issue and then automatically create a new pod to maintain the desired state. Kubernetes provides access to a range of other controllers to similarly manage a range of tasks that collectively ensure reliability.
There’s also a kubebuilder tool that provides the framework IT teams need to construct their own Kubernetes APIs and controllers, including the ability to automatically generate much of the repetitive code required.
Interest in extending the reach of Kubernetes is mainly being driven by the need to reduce costs. As organizations deploy modern cloud-native applications they find themselves managing them alongside legacy monolithic applications that have separate control planes. Each control plane requires a separate IT management team to master that environment. Given the fact that labor remains the single biggest cost of IT, an effort to streamline the number of control planes requires creates a significant opportunity to reduce the total cost of IT.
It’s not clear how quickly organizations are moving to centrally manage control planes. IT leaders will need to have sufficient political capital to overcome resistance to centralizing the management of control planes. IT teams generally prefer the control plane they know best, so convincing them to transition to another will require a lot of fortitude. Far too many IT personnel still define their value by the certification they have achieved to manage specific platforms rather than the function they provide to their organization. Savvy IT leaders will put the issue to their staff to start a discussion that, hopefully, should lead to an outcome based more on reason than emotion.
Ultimately, it’s only a matter of time before the control plane issue is forced. As organizations continue to deploy workloads across multiple clouds and on-premises IT environments that include edge computing platforms, it becomes much less practical to use multiple control planes.
In the meantime, IT leaders should keep track of the number of Kubernetes clusters being deployed. Each additional cluster is essentially pushing IT teams closer toward what will inevitably become an unsustainable crisis. Each IT leader will need to decide for themselves how far in advance they will want to get ahead of the issue but as is the case with any issue the longer it takes to address the bigger the problem it later becomes so, as always, an ounce of prevention applied today will later on be worth multiple pounds of cure.