Kubernetes#
Kubernetes is the operator’s standard control plane for running containers at fleet scale. A cluster reconciles a declared desired state, expressed in YAML against a REST API, into running workloads on a pool of nodes. The orchestrator schedules pods, restarts failed ones, rolls updates forward and backward, routes traffic, and exposes every change through audit logs the operator can read.
For the operator, Kubernetes is force multiplier and target in equal measure. It deploys defended estates and campaign infrastructure without manual touch, but every kube-apiserver exposed on the network, every misconfigured RBAC rule, and every container that escapes its namespace is the next intruder’s pivot. Knowing the object model and the cluster’s component layout is the price of entry to either side.
The pages below break the topic into four parts, the cluster’s internal architecture, the object model the operator writes against, how a cluster gets stood up on the major platforms, and how a real project is built on top of one.
Control plane and data plane. The components a cluster runs.
The object model. Pods, Deployments, Services, ConfigMaps, Secrets, and the rest.
Standing a cluster up. Laptop, AWS, GCP, Azure, vSphere, bare metal.
Building real systems on Kubernetes. The runtime is the means, not the product.