How to Set Up a DevOps Home Lab in 2026 (Free and Cheap Options)

Primary keyword: DevOps home lab setup Secondary keywords: Kubernetes home lab, DevOps lab environment, learn DevOps at home

Introduction

Hands-on practice is what builds real DevOps skill, and you need an environment to practice in. The good news: in 2026, you don’t need expensive hardware or a large cloud bill to build a capable lab. Between local Kubernetes clusters, cloud free tiers, and low-cost VPS options, you can run a meaningful DevOps environment for $10-20 per month — or even less. This guide covers the options, what to run on each, and how to structure your learning.

Option 1: Your Laptop — Free and Surprisingly Capable

For the first few months of learning, your laptop is enough. Modern machines run Kubernetes clusters locally with minimal friction.

Local Kubernetes: minikube or kind

minikube — the most beginner-friendly local Kubernetes option. Single command to start, built-in dashboard, add-ons for common tools.

# Install minikube
brew install minikube   # macOS
# or download from minikube.sigs.k8s.io

# Start a cluster
minikube start --cpus=4 --memory=8192

# Enable useful add-ons
minikube addons enable metrics-server
minikube addons enable ingress

# Access the dashboard
minikube dashboard

kind (Kubernetes in Docker) — runs Kubernetes nodes as Docker containers. Faster to create and destroy, better for testing multi-node setups.

# Install kind
brew install kind   # macOS

# Create a single-node cluster
kind create cluster --name dev

# Create a multi-node cluster
cat <<EOF | kind create cluster --config=-
kind: Cluster
apiVersion: kind.x-k8s.io/v1alpha4
nodes:
- role: control-plane
- role: worker
- role: worker
EOF

What to run locally:

Deploy containerized applications (your own projects)
Practice kubectl commands until they’re reflexive
Set up ArgoCD and practice GitOps
Deploy Prometheus and Grafana and build dashboards
Practice Helm chart installation and customization

Requirements: 8GB RAM minimum (16GB recommended for running multiple services), 4 CPU cores, 20GB free disk.

Docker Desktop

If you haven’t used Docker in production, Docker Desktop is the quickest way to start. It includes a single-node Kubernetes cluster you can enable with one click. Good for initial learning, but minikube or kind gives you more control.

Option 2: Cloud Free Tiers — Real Cloud Without a Bill

Every major cloud provider offers a free tier that’s usable for learning.

AWS Free Tier

EC2 t2.micro — 750 hours/month free for 12 months. Always-free tier afterwards for t2/t3.micro in limited circumstances.
S3 — 5 GB storage free
RDS — 750 hours of db.t2.micro free for 12 months
Lambda — 1 million free invocations per month (permanent)

Best use: Practice Terraform by provisioning and destroying real AWS infrastructure. The free tier EC2 is small, but it’s real Linux in the cloud — perfect for practicing SSH, firewall rules, and basic server administration.

# Set up AWS CLI
aws configure   # enter your access key and secret

# Verify it works
aws ec2 describe-instances --region us-east-1

Azure Free Account

$200 in credits for 30 days
12 months of free services: B1s VMs, 64 GB managed disks, 15 GB blob storage
Always-free tier: App Service (10 web apps), Azure Functions (1M requests/month)

Best use: Terraform practice on Azure, AKS (Azure Kubernetes Service) with the free tier VMs. Good if you’re targeting Azure-heavy employers.

Google Cloud Free Tier

$300 in credits for 90 days
Always-free tier includes an e2-micro VM in us-central1

Best use: GKE (Google Kubernetes Engine) has a free control plane tier, making it the most affordable managed Kubernetes for learning.

Option 3: Low-Cost VPS — $5–20/Month for a Persistent Environment

If you want a lab that’s always running without cloud complexity, a VPS is the right choice.

Recommended Providers

Hetzner Cloud — best value. A CX21 (2 vCPU, 4GB RAM, 40GB SSD) costs ~€4/month. European provider, excellent uptime, simple pricing. Best for Linux and k3s practice.

DigitalOcean — $6/month for 1GB RAM droplet, $12 for 2GB. Good documentation, simple interface. Ideal for beginners.

Linode (Akamai) — comparable to DigitalOcean in pricing and features.

What to Run on a VPS

k3s — a lightweight Kubernetes distribution that runs on a single low-spec VM.

# Install k3s on a fresh Ubuntu VM
curl -sfL https://get.k3s.io | sh -

# Get the kubeconfig
cat /etc/rancher/k3s/k3s.yaml

# Copy to your laptop
scp user@your-vps:/etc/rancher/k3s/k3s.yaml ~/.kube/config
# Update the server IP in the config to your VPS's public IP

A 2GB VPS running k3s can comfortably host: a web application, Prometheus and Grafana, Traefik for ingress, and a small database. That’s enough for meaningful practice.

Option 4: Multi-Node Home Server — For Serious Labs

If you want a more realistic cluster with multiple nodes, Raspberry Pis or refurbished mini-PCs are cost-effective options.

Raspberry Pi Cluster

4 Raspberry Pi 4 (4GB RAM each) costs ~$300-400. Run k3s in a proper multi-node setup — one control plane, three workers. This gives you a real cluster where node failures, pod scheduling, and network policies behave like they would in production.

Refurbished Mini PCs

Intel NUC or Beelink mini PCs (often $80-150 refurbished) run full-featured Kubernetes. A 3-node cluster with 8-16GB RAM each handles serious workloads — Istio, multiple applications, full observability stack.

What to Run in Your Lab (In Order)

Start simple and add complexity as your skills grow.

Stage 1 — Local Kubernetes Basics (weeks 1-4):

Deploy a containerized app you wrote
Practice kubectl CRUD operations for every resource type
Set up Helm, install a chart (e.g., nginx-ingress)

Stage 2 — GitOps and CI/CD (weeks 5-8):

Install ArgoCD, connect it to a GitHub repo
Set up a GitHub Actions pipeline that pushes to your registry
Let ArgoCD deploy automatically when the image tag changes

Stage 3 — Observability (weeks 9-12):

Deploy Prometheus using the kube-prometheus-stack Helm chart
Build Grafana dashboards for your application
Set up an alert rule and trigger it intentionally

Stage 4 — Infrastructure as Code (weeks 13-16):

Provision your cloud VPS with Terraform
Manage all your lab configuration in Git
Practice Terraform state management with remote backends

Stage 5 — Security and Advanced (ongoing):

Set up NetworkPolicies — deny all, allow explicitly
Add cert-manager for automatic TLS certificates
Add a secret manager (Vault or External Secrets Operator)
Practice disaster recovery: delete the cluster, restore from state

Lab Project Ideas

Project: Complete deployment platform — Build a mini version of what a platform team would build: ArgoCD managing apps, Prometheus monitoring them, a Grafana dashboard showing health, cert-manager handling TLS. Document it thoroughly.

Project: Multi-environment setup — Run dev and staging environments on the same cluster using namespaces. Automate promotion from dev to staging via ArgoCD.

Project: Chaos engineering — Intentionally kill pods, nodes, and services. Observe how your monitoring detects it and how Kubernetes recovers. This is how production engineers think.

Conclusion

The lab that will teach you the most isn’t the most elaborate one — it’s the one you actually use. Start with minikube on your laptop, deploy real applications, and add complexity incrementally. A few hours a week of hands-on practice in a real Kubernetes environment will build more skill than any amount of video-watching. Set up your lab this week, pick a project from the list above, and start building.

Want structured lab exercises designed to build production-grade DevOps skills at every phase? The full curriculum is at ashoklabs.com.

Explore the courses →