Summary
K3s was being considered for resource-constrained edge sites (Raspberry Pi/mini-PC) instead of taking Rancher's marketing at face value, so real resource consumption, startup time, and workload performance were compared against K8s on limited hardware.
System Context
- •Test hardware: 4-core / 8GB RAM mini-PC, matching the intended edge deployment
- •Compared against kubeadm K8s 1.29 on identical hardware
- •Benchmark workload: stress-cpu plus a sample Node.js app
Checklist
- ☑Measured idle RAM: K3s vs kubeadm K8s on the same hardware
- ☑Measured time from install to cluster Ready
- ☑Ran a load test workload, compared CPU throttling
- ☑Identified which enterprise K8s features are missing in K3s (e.g. some CSI drivers)
Commands
Step 1
curl -sfL https://get.k3s.io | sh -Step 2
kubectl run stress --image=polinux/stress -- stress --cpu 2Step 3
kubectl top nodesStep 4
k3s kubectl get --raw /metricsConclusion
K3s uses 40% less memory than kubeadm-based clusters on identical hardware. Startup time is under 30 seconds. Suitable for edge deployments and development environments, but lacks some enterprise features.