Advanced BPF kernel features for the container age
summary
Covers recent Linux kernel BPF features and Cilium extensions used in Kubernetes to improve throughput, latency and reliability. Describes using the earliest departure time (EDT) model for egress bandwidth management, new BPF redirection helpers for low-latency switches into Pod network namespaces, and a BPF-based Maglev consistent-hashing implementation used by Cilium’s XDP north‑south load balancer and kube-proxy replacement.
AI-written summary · about
related talks
05mLightning Talk: Back to the Future with eBPF
31mUnleashing the Tides of Kubernetes Networking by Removing kube-proxy
07mLightning Talk: Taming Noisy Network Neigbours with Cilium and eBPF
35mLiberating Kubernetes From Kube-proxy and Iptables
32meBPF: BPF kernel infrastructure
38mUnderstanding and Troubleshooting the eBPF Datapath in Cilium
34mWhat's new in the BPF world?
28mLessons from Building Scalable Network Policy Enforcement with eBPF