Harnessing Observability for 5G Performance: eBPF and OpenTelemetry I...
summary
Shows how eBPF and OpenTelemetry tie into a Kubernetes-native pipeline: bpfman as a CNCF guardian for loaded eBPF programs, the OpenTelemetry Collector ingesting CNF metrics, and export to Prometheus/Grafana for visualization. Demonstrates a sandboxed Open5GS 5G lab sending real AMF/SMF/UPF metrics, an agentic workflow (diagnostic → planning → resolution → validation) that runs remediation via Ansible playbook, and a real failure where a privileged daemonset on a 1500-node OpenShift cluster preempts management pods requiring node-by-node recovery.
AI-written summary · about
chapters (16)
- 0:00Introduction
- 1:24Observability in telecom
- 3:51Traditional observability limits
- 4:54Service mesh and eBPF
- 6:18eBPF misuse risks
- 8:03Red Hat observability tools
- 9:06Adopting observability challenges
- 10:10Telecom data pipelines
- 11:38Enterprise observability experiments
- 12:005G sandbox and OpenTelemetry
- 13:04Agentic AI for networking
- 15:35Anomaly detection demo
- 17:27AI data lifecycle limits
- 18:54Telecom AI use cases
- 20:19Observability insights in practice
- 21:44Closing and Q&A
related talks
11mLightning Talk: 5G CNFs Observability Based on eBPF
34mNetwork Observability with eBPF and OpenTelemetry
30mAgentless instrumentation with eBPF, Prometheus and OpenTelemetry
31mBridging Dev and Ops with eBPF: Extending Observability Upwards and Downwards
36mFrom Monitoring to Observability: eBPF Chaos - Michael Friedrich
27mStep By Step Kubernetes Observability with eBPF
27mObservability Brings Clarity in 5G World
33mEffortless, Standardised Homelab Observability With eBPF