Kubernetes Networking Horror Stories: Issues, Debugging, and Best Practices - Philip Schmid

This talk addresses the challenges operators face in diagnosing and resolving Kubernetes networking problems, including connectivity issues and scalability challenges. The speaker shares practical debugging techniques drawn from extensive real-world Cilium support cases, covering common failure patterns such as running out of IPs or BPF map memory. The session delves into the use of Cilium and Hubble for enhanced network visibility and metrics, along with essential insights into Cilium internals like policy enforcement and communication patterns. Attendees will gain valuable knowledge to better identify issues and configure Cilium, ultimately leading to a more resilient Kubernetes network. The speaker, Philip Schmid, brings a rich background in network engineering and container technologies, with a focused transition to Kubernetes and cloud-native solutions.