[reference] — from roadmap.sh and vendor docs.

Kubernetes (previous note) won the orchestration war, but it isn’t the only option, and “run raw upstream k8s yourself” is rarely the right choice. This note maps the alternatives and — more usefully — the managed options that most teams actually use.

Self-managed alternatives to raw Kubernetes

ToolPosition
Docker SwarmDocker’s built-in orchestrator. Far simpler than k8s (docker service commands, compose-like), good enough for small clusters — but a shrinking ecosystem. Choose it when k8s is overkill and the team already knows Docker.
HashiCorp NomadA simpler, general-purpose scheduler that orchestrates containers and non-container workloads (raw binaries, VMs). Lighter than k8s; pairs with Consul/Vault. Niche but loved where k8s complexity isn’t wanted.
k3s / kind / minikubeLightweight/local Kubernetes distributions. k3s for edge/IoT and small prod; kind and minikube for local development and CI. Same k8s API, less overhead — the right way to learn k8s hands-on.

Managed Kubernetes — what most teams use

Running the control plane (etcd, API server, upgrades, HA) yourself is a serious operational burden. The cloud providers run it for you, and you just manage workloads:

ServiceProvider
EKS (Elastic Kubernetes Service)AWS
GKE (Google Kubernetes Engine)GCP — generally regarded as the most polished; Google created k8s
AKS (Azure Kubernetes Service)Azure

Managed k8s removes the hardest, most dangerous part (control-plane operation and upgrades) while keeping the standard k8s API — so your YAML and tooling are portable across all three. This portability is the quiet superpower of standardizing on k8s: the same manifests run on any provider or on-prem, which is the main hedge against cloud lock-in.

OpenShift — the enterprise distribution

Red Hat OpenShift is Kubernetes plus an opinionated enterprise layer: built-in CI/CD, an integrated container registry, stricter security defaults (it won’t let containers run as root by default), a developer console, and commercial support. It’s k8s underneath, so skills transfer, but it adds guardrails and batteries that large regulated organizations want. Common in enterprises already on Red Hat — which connects to the RHEL material in the RHCSA track.

How to choose

  • Learning / local dev → kind, minikube, or k3s.
  • Small team, simple needs, already on Docker → Docker Swarm (or honestly, a managed platform / serverless containers to skip orchestration entirely).
  • Production at real scale on a cloud → managed k8s (GKE/EKS/AKS) — don’t run the control plane yourself unless you have a strong reason.
  • Large regulated enterprise wanting a supported, opinionated platform → OpenShift.
  • Mixed container + non-container workloads, wanting simplicity → Nomad.

The meta-point: the interesting decision is rarely “which orchestrator” — it’s “do I need an orchestrator at all, or would managed containers / serverless / a PaaS deliver this with far less operational surface?” Reach for Kubernetes when you genuinely have the scale and complexity that justify it, not by default.

  • Kubernetes — the standard the rest of this landscape is measured against
  • Cloud — where managed k8s runs
  • Serverless — the “skip orchestration” alternative
↑ Back to 05 — Container Orchestration