Kubernetes Roadmap

Zero to production

The complete free Kubernetes roadmap, from your first pod to operators, storage, troubleshooting, and production operations. Built from our own course, workshops, articles, and labs.

8 stages · 43 resources ready · 8 in the works, all free · every stage ends with a checkpoint you can verify on a real cluster

Your journey: 43 steps. Check them off as you go, this page remembers.

Foundations

~1-2 weeks

Everything else builds on a mental model of the control plane. Start here even if you have used kubectl before.

✓ Checkpoint: you can now

Explain to a colleague what the control plane does, and why pods (not containers) are the unit of scheduling.

⚡ Prove it

Draw the architecture from memory: API server, etcd, scheduler, kubelet, and what talks to what.

Run it locally

~a weekend

Theory sticks when you break things. Get a cluster you can destroy without fear.

✓ Checkpoint: you can now

Spin up and destroy multi-node clusters without fear.

⚡ Prove it

Create a 3-node cluster, kill a node, and watch what happens to its pods.

Workloads & services

~2-3 weeks

Now that you have a cluster, learn how apps actually run on it and how traffic reaches them.

Autoscaling: HPA, VPA, KEDA, and when each appliesguide coming
Progressive delivery: rolling, blue-green, and canary releasesguide coming
Cluster essentials: cert-manager, external-dns, and the add-ons every cluster runsguide coming

✓ Checkpoint: you can now

Deploy a stateful app with services in front and explain the full traffic path.

⚡ Prove it

Deploy an app plus a database, expose it via ClusterIP, NodePort, and Ingress, and explain the difference.

Internals

~2-3 weeks

You can use Kubernetes; now learn what it does under the hood, because that is what debugging demands.

✓ Checkpoint: you can now

Answer 'what happens when I kubectl apply' end to end, from API server to running container.

⚡ Prove it

Follow one pod from apply to Running: watch the events, the scheduler decision, and the kubelet start.

Storage & secrets

~1-2 weeks

Stateless is easy. Real systems have state and credentials, and this is where clusters get hurt.

CSI from scratch: how volumes really attach (article)guide coming

✓ Checkpoint: you can now

Choose the right secrets approach for a team and defend the choice.

⚡ Prove it

Wire External Secrets or Sealed Secrets into a demo app end to end.

Operators & CRDs

~2-4 weeks

Kubernetes' real power is extending the API itself. This is how the ecosystem you use every day is built.

✓ Checkpoint: you can now

Read any operator's CRDs and predict what its reconcile loop will do.

⚡ Prove it

Build a tiny operator with Kubebuilder that labels namespaces, and break its reconcile loop on purpose.

Security

~2-3 weeks

With workloads running, lock them down. Everything here comes up in real audits and real incidents.

✓ Checkpoint: you can now

Harden a cluster with RBAC, network policies, and admission control, and know which audit logs matter.

⚡ Prove it

Lock down a namespace so a compromised pod can reach nothing except its own database.

Production operations

~ongoing

The final stretch: keeping clusters alive under pressure. Upgrades, observability, chaos, and 3am debugging.

Written troubleshooting playbook: common failures, fast diagnosisguide coming
Multi-cluster and cell-based architectures: how large platforms scaleguide coming
Backup and disaster recovery that actually restoresguide coming

✓ Checkpoint: you can now

Debug a broken cluster under pressure and upgrade one without downtime.

⚡ Prove it

Break your local cluster on purpose (kill etcd, wedge a node) and recover it.

Want to know when the 🚧 steps ship?

New articles, videos, and labs land in the newsletter first.

Get updates →