CF Networking: All Your Packets are Belong to Us
summary
Cloud Foundry networking introduces per-container IPs and app-level allowlists, implemented via the CNI standard and a built-in silk CNI plugin, a central policy server with a policy agent, and VXLAN overlay tunnels. Garden calls an external networker that assigns IPs, programs iptables on the destination, and encapsulates packets over the VXLAN overlay so container-to-container TCP works without the gorouter; demo shows policy enforcement returning a cat image. Service discovery (Eureka/bring‑your‑own) remains required for load balancing and name resolution.
AI-written summary · about
chapters (8)
- 0:00Introduction and team
- 1:05Networking challenges overview
- 3:54CF Networking 1.0 release
- 9:10Architecture deep dive
- 16:47Data plane and packet flow
- 17:59Live demo of networking
- 21:13Future roadmap and feedback
- 26:34Q&A session
related talks
30mCF Networking: All Your Packets Belong to Us - Usha Ramachandran & Angela Chin, Pivotal
29mContainer Networking in Cloud Foundry
31mCloud Foundry Networking: Enabling Direct Communication for Microservices - Chin, McClure
01h 05mCloud Foundry Networking with VMware NSX - Usha Ramachandran, Sai Chaitanya
06mKeynote: Container to Container Networking for Cloud Foundry and Kubernetes
26mContainer Networking: The Play of BPF and Network NS with different...
39mThe State of the Cloud Native Network: Deep-dive from TAG-Network
40mCloud, Kubernetes, and Service Networking - Taming the Turtles
