Gateway API: Basics, Nuance, and the Real World
summary
Gateway API replaces Ingress by splitting responsibilities into GatewayClass/Gateway/HTTPRoute so infra teams provision listeners and app teams own routes; controllers and CRDs install separately; demo uses kind with a Traffic controller and nginx, running Siege, showing rolling updates produce 502s because endpoints drop early. Solution demos Kubernetes alpha pod lifecycle sleep action to delay endpoint removal and achieve zero‑downtime, and a cross‑namespace referenceGrant pattern for secret access; it works in the demo.
AI-written summary · about
chapters (12)
- 0:00Introduction and speaker background
- 1:27Why Gateway API matters in Kubernetes
- 4:58Kubernetes networking model and Ingress limitations
- 15:29Problems with Ingress and the need for Gateway API
- 18:19Gateway API basics and resource model
- 23:41Gateway classes and infrastructure provisioning
- 35:33Gateway API nuances and multi-implementer challenges
- 37:17Reference Grant for cross-namespace access
- 43:03Live demo of Gateway API in action
- 48:07Zero downtime deployments and pod shutdown issues
- 55:12Pod lifecycle sleep action feature
- 57:38Summary and recap of Gateway API benefits
related talks
35mGateway API: The Illustrated Guide to the Future of Kubernetes Networking
28mBeyond Kubernetes Ingress (with Gateway API)
32mGateway API: A New Set of Kubernetes APIs for Advanced Traffic Routing
26mGetting Hands-On with the New Kubernetes Gateway API
47mGatewayAPI, 10 Years of Maturation for a New Kubernetes API
29mGateway API: The new Kubernetes Ingress that Unifies Load Balancing and Network Configs
36mGateway API The new way to travel north south
50mGateway API, 10 Years of Maturation for a New Kubernetes API