Making Kubernetes Networking Ready for Service Providers
summary
Kubernetes lacks telco-ready ingress/egress and protocol support, so F5 builds a service proxy (F5 SPK) that replaces the built-in service proxy with CRDs running entirely in-cluster, exposes BGP, egress NAT, IPv4↔IPv6 translation, and a layer‑4 firewall, and supports TCP/UDP/SCTP, Diameter and GTP; production deployment with a tier‑one carrier shows it ties ingress and egress to present a stable CNF endpoint despite pod churn. Roadmap targets Gateway API alignment and CNI‑independence so other gateways can interoperate.
AI-written summary · about
chapters (10)
- 0:00Kubernetes in telecom networks
- 1:04Kubernetes networking limits
- 3:09Gaps for service providers
- 8:08Extending vs breaking patterns
- 9:13Anti-patterns in CNF networking
- 13:30Service proxy concept
- 16:22CNF protocols and security
- 19:54Centralizing telecom security
- 21:39Standardizing CNF testing
- 23:26Q&A on IP mapping
related talks
10mLightning Talk: Kubernetes Gateway API for Complex Environments and Service Providers
01mLearn about the latest Service Provider Cloud-native Network Functions
28mProxy Service: A New Network Traffic Abstraction in Kubernetes
15mSimplify Kubernetes Connectivity
09mLightning Talk: Hacking Kubernetes Networking to Create Telco Solutions
34mUnderstanding the Kubernetes Networking Layer
31mUnleashing the Tides of Kubernetes Networking by Removing kube-proxy
35mLiberating Kubernetes From Kube-proxy and Iptables