K8s

Container Orchestration

Kubernetes Learning Roadmap

Kubernetes (K8s) is the industry-standard system for deploying, scaling, and managing containerized applications in production. If your company runs microservices at scale, it almost certainly runs Kubernetes. K8s expertise is one of the most in-demand and highest-paying skills in DevOps and platform engineering.

Total time

4–6 months

Stages

6 stages

Job-ready at

Stage 4

Career paths

3 paths

Before you start — Prerequisites

✓

Docker and containers (required)

Kubernetes orchestrates containers — you must understand Docker images, containers, and registries first.

✓

Linux command line

kubectl commands run in a shell, and most K8s debugging happens via Linux tools.

✓

Basic networking (IP, DNS, TCP)

Kubernetes networking (Services, Ingress, DNS) requires networking fundamentals.

✓

YAML syntax

All Kubernetes resources are defined in YAML manifests.

Stage 1: Core Concepts

Pods, Deployments, and the control plane

2–3 weeks

Topics to cover

Cluster architecture: control plane vs worker nodesPods: the smallest deployable unitkubectl: apply, get, describe, delete, logsDeployments and ReplicaSetsServices: ClusterIP, NodePort, LoadBalancerNamespaces for resource isolationYAML manifests: apiVersion, kind, metadata, specLabels, selectors, and annotationsminikube or kind for local clusterskubectl contexts and kubeconfig

Bold = especially important

Milestone

You deploy a containerized app to a local K8s cluster, expose it via a Service, scale it up and down, and read pod logs.

Project to build

Deploy a web app to minikube

Write Deployment and Service manifests for a containerized web app, apply with kubectl, expose on localhost, scale to 3 replicas.

Stage 2: Configuration & Storage

ConfigMaps, Secrets, and persistent volumes

2–3 weeks

Topics to cover

ConfigMaps: externalize app configurationSecrets: base64-encoded sensitive valuesMounting configs as env vars or filesPersistentVolumes (PV) and PersistentVolumeClaims (PVC)StorageClasses and dynamic provisioningStatefulSets for stateful workloadsInit containersResource requests and limits (CPU/memory)Probes: liveness, readiness, startupPod disruption budgets

Bold = especially important

Milestone

You configure apps with ConfigMaps and Secrets, attach persistent storage, and set resource limits and health probes correctly.

Project to build

Stateful database deployment

Deploy PostgreSQL as a StatefulSet with a PVC, ConfigMap for config, Secret for credentials, readiness probe.

Stage 3: Networking & Ingress

Route external traffic into your cluster

2–3 weeks

Topics to cover

Cluster DNS: how pods find each otherIngress resources and Ingress controllersNGINX Ingress controller setupTLS termination with cert-managerNetwork Policies: firewall between podsDNS debugging: nslookup and dig in podsHeadless services for StatefulSetsExternalDNS for automatic DNS recordsGateway API (next-gen Ingress)Service mesh basics: Istio or Linkerd concepts

Bold = especially important

Milestone

You route external HTTPS traffic to multiple services in your cluster via Ingress, with TLS certificates managed automatically.

Project to build

Multi-service Ingress with TLS

Deploy two apps, route traffic by hostname via NGINX Ingress, add cert-manager for automatic Let's Encrypt certs.

Stage 4: Scaling, Updates & RBAC

✓ Min. Job Ready

Production reliability and access control

3–4 weeks

Topics to cover

Horizontal Pod Autoscaler (HPA)Vertical Pod Autoscaler (VPA)Rolling updates and rollbacksDeployment strategies: RollingUpdate vs RecreateRBAC: Roles, ClusterRoles, RoleBindingsServiceAccounts for pod identityNode affinity and anti-affinityTaints and tolerationsDaemonSets for node-level agentsJobs and CronJobs

Bold = especially important

Milestone

You configure HPA for auto-scaling, execute zero-downtime rolling deployments, and set up RBAC for multi-team cluster access.

Project to build

Auto-scaled app with RBAC

HPA on CPU metric, rolling update strategy, RBAC for developer and ops roles, CronJob for scheduled cleanup.

Stage 5: Observability & Security

Know what your cluster is doing

2–3 weeks

Topics to cover

Prometheus for metrics collectionGrafana dashboardskube-state-metrics and node-exporterLoki or Elasticsearch for log aggregationPod Security Standards (replace PSP)OPA Gatekeeper / Kyverno for policySecrets management with External Secrets OperatorCluster auditing with Falcokubectl top and resource usage analysisetcd backup and cluster disaster recovery

Bold = especially important

Milestone

You run Prometheus + Grafana for metrics, aggregate logs, and apply Pod Security Standards. You can recover from a control plane failure.

Project to build

Full observability stack

kube-prometheus-stack (Prometheus + Grafana + Alertmanager) deployed via Helm, custom dashboard for app metrics.

Stage 6: Helm, GitOps & CKA Prep

Production toolchain and certification

Ongoing

Topics to cover

📦 Helm charts: templating K8s manifests🔄 GitOps: ArgoCD or Flux CD🏗️ Cluster bootstrapping: kubeadm📜 CKA exam preparation🔌 Custom controllers and operators📡 Service mesh: Istio traffic management🌐 Multi-cluster: Cluster API, Admiral⚡ eBPF networking: Cilium, Hubble

Bold = especially important

Milestone

You deploy apps via Helm and manage your cluster declaratively with GitOps. You are ready for the CKA exam.

What to learn next with Kubernetes

Skills that pair directly with Kubernetes in job postings

AWS

EKS (Elastic Kubernetes Service) is the managed K8s on AWS

Docker

Kubernetes runs Docker containers — deepen your container knowledge

Linux

Node debugging, kernel tuning, and cgroups are Linux-level concerns

Kubernetes and its ecosystem are written in Go — build controllers and operators

Git

GitOps with ArgoCD/Flux manages K8s configs from a Git repository

Career paths that use Kubernetes

DevOps Engineer Platform Engineer Site Reliability Engineer

Frequently asked questions

Do I need to learn Docker before Kubernetes?

Yes — absolutely. Kubernetes orchestrates containers. If you don't understand what a container image is, how networking inside containers works, and how to write a Dockerfile, Kubernetes will be confusing. Complete at least the first 3 stages of the Docker roadmap first.

What is the CKA certification and should I get it?

The Certified Kubernetes Administrator (CKA) is the most respected K8s credential. It's a hands-on, performance-based exam (not multiple choice) — you solve real cluster problems in a terminal. Most serious K8s job applicants pursue it. Budget 3–4 months of study after finishing this roadmap.

How do I practice Kubernetes without a cloud account?

Use minikube (local single-node cluster), kind (Kubernetes in Docker), or k3s (lightweight K8s for Raspberry Pi/VMs). All are free and run on your laptop. For managed clusters, GKE has a free trial and the GKE Autopilot free tier is the cheapest managed option.

Kubernetes vs Docker Swarm — should I learn Swarm?

No. Kubernetes won. Docker Swarm still exists but is rarely used for new projects. Your time is better spent on Kubernetes, which is the industry standard for production container orchestration.

Is Kubernetes overkill for small projects?

Yes — Kubernetes adds significant operational complexity. For a side project or a small app with one or two services, Docker Compose or a platform like Fly.io is more appropriate. K8s pays off at scale: many services, many replicas, or strict uptime requirements.

Ready to start learning?

Browse our curated Kubernetes resources — every course, book, and video ranked by quality.

View Kubernetes Resources →