The Problem Helm Solves

From a pile of YAML to a package

A real application on Kubernetes is not one file. A typical web service needs a Deployment, a Service, an Ingress, a ConfigMap, a ServiceAccount, maybe a HorizontalPodAutoscaler and a PodDisruptionBudget. Now repeat that for dev, staging and production, where only a few values differ: replica count, image tag, hostnames, resource limits. Copy-pasting YAML per environment drifts quickly. And kubectl apply has no idea which objects belong together, so removing an app or going back to last week's version is manual. Helm is the package manager for Kubernetes. It bundles templated manifests into a chart, fills them in with values for each environment, installs them as a named release, records every revision, and can upgrade, roll back or uninstall the whole set as one unit. Helm is a graduated project of the Cloud Native Computing Foundation.

Chart plus values becomes a release

The same chart, combined with different values, produces a release per environment.

What Helm replaces

Without Helm you manage each file and each environment by hand; with Helm one command installs or upgrades the whole set.

# without Helm: apply many files, per environment, and track them yourself
kubectl apply -f deployment.yaml -f service.yaml -f ingress.yaml -f configmap.yaml

# with Helm: one release, one command, with history
helm install shop-api ./charts/shop-api -f values-prod.yaml --namespace shop
helm history shop-api --namespace shop
helm rollback shop-api 3 --namespace shop

apt or npm for your cluster

Just as apt install nginx or npm install express fetches a package and its dependencies, helm install fetches a chart and installs everything it needs into the cluster, and remembers what it installed so it can upgrade or remove it later.

Quick check: Which capability does Helm add that plain `kubectl apply` of files lacks?

• Tracking a set of objects as one versioned release that can be rolled back
• Running containers
• Scheduling pods onto nodes
• Building container images