Helm + K8s Distributions
20 min
Day 72: Vanilla vs K3s vs RKE2 vs managed
Vanilla Kubernetes vs K3s vs RKE2 vs managed — an honest comparison
You rarely install "Kubernetes" as one monolithic thing — you choose a distribution, each making different trade-offs.
- Vanilla (kubeadm) — full control, full complexity; you manage every component yourself. Good for learning internals deeply, painful to operate at scale unassisted.
- K3s — a genuinely lightweight distribution (single binary, SQLite by default instead of etcd for small clusters), built for edge/IoT/dev — you can have a working cluster in under a minute.
- RKE2 — Rancher's security-hardened, CIS-benchmark-compliant distribution, popular in regulated/enterprise on-prem environments.
- Managed (EKS, GKE, AKS) — the cloud provider runs and patches the control plane for you; you only manage worker nodes and workloads. Highest convenience, less control, and a real cost.
The real decision driver
This choice is rarely about raw capability — it's about who you want operating the control plane. Learning on vanilla/K3s builds the deepest understanding; running production on managed K8s (Phase 20's EKS) is what most teams actually do, precisely because someone else absorbs the control-plane operational burden.
Key terms
- K3s
- A lightweight, single-binary Kubernetes distribution suited for dev/edge/IoT.
- Managed Kubernetes
- A cloud provider's Kubernetes offering where the control plane is operated on your behalf.
A startup with a small platform team is choosing between self-managed vanilla Kubernetes and EKS (managed). What's the strongest argument for EKS?
