Kubernetes and the CNI: Where We Are and What's Next - Casey Callendrello, CoreOS
summary
CNI is a vendor‑neutral protocol plus a simple on‑disk binary and a long‑running daemon: the binary creates a veth pair, allocates the pod IP and the daemon programs cluster reachability (overlay, BGP/OSPF, or cloud route APIs). Kubelet scans one config file every 5s so only write one; deploy plugins self‑hosted, run daemons in host network, store volatile state in /run, and handle hairpin and GC. CNI 0.4 adds result caching and more capability arguments and plugin chaining enables lightweight extensions.
AI-written summary · about
chapters (9)
- 0:00Introduction to CNI
- 1:07CNI Protocol Design
- 3:33CNI in Kubernetes
- 6:13Networking challenges
- 9:06Plugin functionality
- 13:56Best practices for operators
- 19:43Best practices for developers
- 31:46Future of CNI
- 35:54Q&A and conclusion
related talks
33mKubernetes Networking: How to Write a CNI Plugin From Scratch
37mIntroduction to CNI, the Container Network Interface Project
31mEverything You Ever Wanted to Know About CNIs (Container Network Interface) and ...
31mMayday! CNI Overboard!
34m"My CNI Plugin Did… What?!": Debugging CNI with Style & Aplomb
31mFrom CNI to Service Mesh: Demystifying Kubernetes Networking
32mDemystifying CNI Writing a CNI from Scratch
45mHow to Build a Kubernetes Networking Solution from Scratch