Before you put it in front of production

Kong Course · lesson 14 of 15 · 4 min read

The gateway becomes the thing everything depends on - here is what that demands.

Open this lesson in the learning hub

Key points

  • Run at least two Kong nodes behind a load balancer. A single gateway is a single point of failure.
  • Never expose the Admin API publicly - bind it to localhost or a private network.
  • Set sensible timeouts and retries per service, or one slow upstream ties up gateway workers.
  • Alert on 4xx and 5xx per route, not in aggregate - one broken route hides in a global average.
  • Rehearse a config rollback before you need one; with decK that is a git revert plus a sync.

Two nodes, a private Admin API, per-route alerting and a rehearsed rollback - that is production ready.

Kong on Kubernetes Where a gateway spends its time

This is a reading copy. The full lesson — with the visual explainer, the interactive lab and a Run button for the code — lives in the Kong Course course, and every lesson in it is listed on the Kong Course contents page.