A good platform makes the right way the easy way: developers get a working, secure environment from a template, deploy by merging to Git, and see their own metrics. Here is how such a platform is put together, and the early decisions that matter most.
Control plane3 nodes
DeliveryGitOps
TenancyNamespace per team
Upgrades~3 Kubernetes releases a year
Reference architecture
How the pieces fit together
Scroll sideways to see the whole diagram →Developers never touch the cluster directly. Code goes through CI, images through the registry, and configuration through Git.
Part
What it does
1 Build
CI builds the container image, runs tests and scans it for known vulnerabilities.
2 Images
Only scanned, signed images from your registry are allowed to run.
3 Configuration from Git
The GitOps controller applies everything the cluster runs from Git, including policies and add-ons.
4 Feedback
Each team sees its own metrics, logs and alerts, so problems are found by the people who can fix them.
Team namespaces
Each team gets a namespace with resource quotas and network policies, created from a template.
Managed or self-run?
Managed Kubernetes (EKS, AKS, GKE, OKE)
Self-run (RKE2, OpenShift, kubeadm)
Control plane
Run and upgraded by the provider
Run and upgraded by you
Good for
Public cloud workloads, small platform teams
On-premises, private cloud, strict data rules
Watch out for
Cloud cost and provider-specific features
Upgrade discipline and on-call ownership
Sizing a first cluster
Control plane: three nodes, so it survives one failure and can be upgraded one node at a time.
Workers: enough capacity for peak load with one node lost, plus room for rolling updates.
Node pools: separate pools for general workloads, memory-heavy workloads and GPUs if needed.
Storage: a CSI driver for your storage (Ceph via Rook, vendor arrays, or cloud volumes) for stateful services.
Tell us how many applications and teams you have, how you deploy today and where it should run. We will come back with an honest view: whether Kubernetes fits, what a lean platform looks like, and what it takes to run.