A High-Schooler’s Guide to Kubernetes Network Observability
Talkat KubeCon + CloudNativeCon North America 2020 Virtual · on Thursday, 2020-11-19· at 14:55· Intrado Virtual Event Platform,·
summary
Kube-netc uses eBPF (Datadog’s eBPF library) to capture per-connection bytes in/out, aggregates those counts in a tracker, enriches each IP by querying the Kubernetes API for pod/service labels, and exposes labeled time-series on a Prometheus endpoint so a supplied Grafana dashboard can show traffic per object. The system deploys as a daemonset, runs on kind, and the implementation is functional — curlable metrics and live dashboards demonstrate the end-to-end pipeline built by a novice contributor.
AI-written summary · about
chapters (17)
- 0:00Intro
- 0:15Who am I?
- 1:03Enter Nirmata
- 2:43The Project: kube-netc
- 4:01The Goal of kube-netc
- 6:03eBPF Crash course
- 8:55Open-Source eBPF Projects
- 10:13Implementing Prometheus
- 10:42Basics of Prometheus
- 12:03The Landscape
- 12:56Grafana Visualization
- 14:22What is Kubernetes?
- 15:51The Daemon Set
- 19:39Great Starter Issues!
- 20:20Map of the Components
- 22:53Demo
- 28:56Key Takeaways
related talks
27mStep By Step Kubernetes Observability with eBPF
42mMonitoring Kubernetes Traffic with eBPF
34mNetwork Observability with eBPF and OpenTelemetry
31mTrue Kubernetes Observability with eBPF
46mNetworking in Kubernetes: can we make it clearer?
41mFrom PIDs to Pods: our journey to connect eBPF and Kubernetes
29m“Are These Things Talking to Each Other?” - Observing Kubernetes Networking...
30mObservability in Kubernetes: Unlocking the Power of eBPF