Skip to content

Kubernetes vs App Platform

Both run containers. They differ in how much of the platform you want to own.

What each product is

App Platform is managed container hosting. You push an image or connect a Git repo; the platform handles deployment, scaling, TLS and DNS. You pay for runtime hours and can scale to zero.

Managed Kubernetes is a real Kubernetes cluster with a WAYSCloud-operated control plane and private worker nodes you size yourself. You bring Helm charts, operators, ingress controllers and GitOps. You pay a cluster fee plus the nodes by the hour.

When to use App Platform

  • One to a handful of services, each a container
  • You want Git-push deploys, rollbacks and automatic TLS without configuration
  • Traffic is uneven and idle cost matters (scale to zero)
  • Nobody on the team wants to learn kubectl

When to use Kubernetes

  • You already deploy with Helm, Kustomize, ArgoCD or Flux
  • You need operators, sidecars, CRDs, network policies or custom ingress
  • You run many services with shared infrastructure (service mesh, queues, cron jobs)
  • You need persistent volumes, StatefulSets or node affinity
  • You want the same cluster API you use elsewhere, with EU data residency

Key differences

App PlatformKubernetes
UnitApp (container + revisions)Cluster (control plane + node pools)
You manageImage and environmentEverything above the node: workloads, ingress, RBAC
ScalingInstances per app, scale to zeroNodes per pool (1–16), HPA/VPA inside the cluster
NetworkingAutomatic routing + TLSLoad balancers on demand, extra IPv4 with reverse DNS, API allow-list
Persistent storageNoYes (encrypted SSD volumes)
BackupsRevisionsDaily cluster + volume backups, restore from dashboard
AccessLogs, metricskubectl, kubeconfig, Cloud Shell
BillingHourly per instanceMonthly cluster fee + hourly nodes + per-use IPs/LB/storage
Time to first deployMinutes~10 minutes to a running cluster, then your tooling

Decision guide

Start with App Platform if your app is a container and you want the platform to disappear.

Start with Kubernetes if you have a cluster-shaped workload today, or a team that already speaks Kubernetes.

The two combine well: run the platform pieces (databases via Databases, object storage via Storage) as managed services and choose per workload whether it lives in an App or in the cluster.

Service documentation