Kubernetes Clusters

Kubernetes is the default Cluster type for production workloads on ModStack. Rather than running our own Kubernetes distribution, ModStack integrates with the major managed-Kubernetes services so you get the operational maturity of those platforms with the developer experience of ModStack on top.

Overview

A Kubernetes Cluster on ModStack is a managed Kubernetes cluster from one of the supported cloud providers. ModStack deploys and operates your workloads on it — rolling upgrades, autoscaling, health probes, secret management, observability — without exposing you to the day-to-day operational surface of Kubernetes itself.

The same Service definitions and Modules that run on a Single Node Cluster or a Local Dev Cluster run unchanged on a Kubernetes Cluster. The cluster type is a runtime substrate concern — not an application concern.

Supported Managed Kubernetes Providers

ModStack integrates with managed Kubernetes services from the major cloud providers. The supported providers today, and those on the near-term roadmap, are:

Amazon EKS Available
Azure AKS Coming soon
Google Kubernetes Engine (GKE) Coming soon
DigitalOcean Kubernetes (DOKS) Coming soon

Hosted or BYOC

Kubernetes Clusters can be run in either of the two Project hosting models:

  • Hosted — ModStack provisions and operates the underlying EKS (or future-supported) cluster on your behalf. You do not need a cloud account; billing flows through ModStack.
  • BYOC — You bring an existing managed Kubernetes cluster from a supported provider. ModStack connects to it, deploys your workloads to it, and operates them, but the cluster itself sits inside your cloud account and your cloud bill.

See Hosted vs BYOC for a fuller comparison of the two models.

What ModStack handles for you

Working with raw Kubernetes is a substantial operational commitment. ModStack handles the parts of that surface that have nothing to do with your application:

  • Rolling deployments and rollbacks of your Services.
  • Health probes, readiness gating, and restart policies.
  • Autoscaling configuration — both pod-level and (where supported) node-level.
  • Ingress, TLS, and service-to-service networking.
  • Secret and config injection.
  • Log, metric, and trace collection from every workload.