The basics of observing Kubernetes: a bird-watcher's perspective, with Miguel Luna
Sep 3, 2024
Miguel Luna, an expert in Observability within Kubernetes, shares his insights on key components like metrics, logs, and traces. He delves into essential tools such as OpenTelemetry and discusses the transformative role of AI in monitoring systems. Listeners will learn about practical steps for implementing observability, improving alert management, and the importance of clear communication among teams. Miguel also emphasizes visual thinking as a powerful tool for navigating complex technical documentation, making observability more accessible.
AI Snips
Chapters
Transcript
Episode notes
From Monolith Releases To 3,000 Deploys
- Miguel moved from telco monoliths to Kubernetes during a CI/CD migration that increased releases from 26 to ~3000 per year.
- He learned Kubernetes by writing an article and presenting at KubeCon, which jump-started his cloud-native career.
Observability Vs Monitoring
- Observability is not the same as monitoring and aims to reveal unknown unknowns about system behavior.
- Metrics, logs, and traces form core signals and a fourth (profiling) is emerging to complete the picture.
Start With SLOs
- Start observability from SLOs and work downward to identify what impacts them.
- Use SLO breaches to guide investigation through services, pods, nodes and underlying hardware.