Panel: Telemetry That Matters
summary
Claims that >50% of metrics are never queried and one customer hit $2M in telemetry overages motivate concrete fixes: use OpenTelemetry collectors and the Kubernetes operator for zero‑code instrumentation, add OBI eBPF for network traces, apply RED metrics and correlated traces+logs+profiles for incidents, enforce semantic conventions, run cardinality explorer and drain/dedup processors, use retroactive sampling and routing connectors to split premium vs cheap storage, and manage fleets with an OpAMP server and OpAMP gateway to prevent config drift at 100,000+ collectors.
AI-written summary · about
chapters (9)
- 0:00Telemetry panel intro
- 0:44More telemetry insightful?
- 3:11Essential signals for incidents
- 6:02Correlating telemetry signals
- 7:50Instrumentation strategies
- 11:41Optimizing telemetry pipelines
- 18:41Telemetry in AI flows
- 23:17Scaling with OpAmp collector
- 26:27Key takeaways and future
related talks
34mPanel: OpenTelemetry: Bridging Platform and Enablement
36mPanel: OpenTelemetry...
11mObservability 2.0 - Much more than just logs,metrics and traces
02h 41mThe Science of Signals: Mastering Telemetry for Observability
34mKubernetes Observability with OpenTelemetry
50mPanel: Observability
27mThe Signal in the Storm: Practical Strategies for Managing Telemetry Overload
49mAdding OpenTelemetry to Production Apps Lessons Learned