Journey Through Kubernetes Ingress: Past, Present, and Beyond
summary
Chronicles the concrete tradeoffs: early patterns like NodePort limits and one load balancer per service hit scale and cost problems, Ingress + controller solves host/path routing but relies on vendor-specific annotations or CRDs, implementations like NGINX/HAProxy/Traefik often bypass kube-proxy for better throughput, and Gateway API (GA Oct 2023) aims for a common schema so multiple data planes interoperate; shows KEDA autoscaling integration, real deployments, and why Gateway API convergence matters now.
AI-written summary · about
chapters (21)
- 0:00Introduction
- 1:05Audience Kubernetes experience
- 2:08Exposing Kubernetes services evolution
- 4:13Worker node networking recap
- 6:03Early external exposure methods
- 8:12NodePort and LoadBalancer limits
- 10:31Advanced routing and service mesh need
- 11:35Ingress and Ingress controller
- 13:22Annotation-based config challenges
- 15:41CRDs for extended capabilities
- 17:13Gateway API introduction
- 19:35Security in Gateway API
- 21:25Gateway API layered architecture
- 22:50EngineX Gateway Fabric
- 24:19Performance bypassing kube-proxy
- 25:05North-South traffic and canary testing
- 25:46Scalability with Gateway API
- 27:16Gateway API future
- 29:27Adopting Gateway API
- 29:51Kubernetes ingress summary
- 30:55Q&A
related talks
28mApplication delivery on k8s - the past, present and future
30mGateway to the Future: A Brief History of Kubernetes Ingress
25mKubernetes Ingress: Do More with Less
15mSimplify Kubernetes Connectivity
20mInside Kubernetes Ingress
32mKubernetes Ingress: Your Router, Your Rules
35mGateway API: The Illustrated Guide to the Future of Kubernetes Networking
43mKubernetes Ingress and API Gateways