Lightning Talk: Taming Noisy Network Neigbours with Cilium and eBPF
summary
Cilium’s Bandwidth Manager injects eBPF into pod interfaces to enforce egress rate limits using an EDT algorithm plus the kernel FQ qdisc; example limit 10 Mbps. Benchmarks show baseline ~4 GB/s per flow and 1 ms latency, enabling the manager drops ingress to ~500 Mbps and raises latency to ~80 ms. Enabling BPF host routing (Cilium 1.9) restores throughput to ~4 GB/s and cuts latency to ~15 ms, but global timestamping in EDT still increases latency.
AI-written summary · about
related talks
44mCilium - Container Security and Networking Using BPF and XDP
10mLightning Talk: Closing the Gap: Fair North–South Bandwidth Management for Tenants
31mUnleashing the Tides of Kubernetes Networking by Removing kube-proxy
38mUnderstanding and Troubleshooting the eBPF Datapath in Cilium
29mSimplifying the Networking and Secur...
11mCilium Service Mesh
46mNetwork Traffic Quality of Service (QoS) Classes for Containers & Pods Via EBPF/XDP
13mLightning Talk: Accelerating Cilium: Our Journey on Offloading Cilium To DPU