netstack.fm OpenTelemetry, eBPF, and Metoro
11 snips
Sep 2, 2025 Chris Batterbee, co-founder of Metoro and experienced software engineer, dives into the fascinating world of observability in modern software. He discusses how Metoro uses eBPF technology for automatic instrumentation, simplifying observability in Kubernetes environments. The conversation explores the integration of OpenTelemetry, common challenges developers face, and the potential of AI in diagnosing system issues. Chris also shares insights from his Y Combinator experience and highlights the competitive landscape of observability tools, positioning Metoro uniquely in the market.
AI Snips
Chapters
Transcript
Episode notes
From Palantir Work To Metoro
- Chris Bett erbee describes building Metoro after seeing observability gaps at Palantir while orchestrating large Spark jobs at scale.
- He used eBPF to auto-instrument apps to avoid heavy manual instrumentation work.
Kernel-Level Auto-Instrumentation
- eBPF inspects kernel syscalls to auto-extract protocol-level data like HTTP, gRPC, Postgres and produce spans without app code changes.
- For TLS it hooks into common libraries to read unencrypted data before encryption.
Why eBPF Needs OpenTelemetry
- eBPF gives per-call spans but struggles to correlate spans across threads and async boundaries.
- OpenTelemetry fills that gap by enabling application-level spans and richer cross-thread trace propagation.