5 reasons why (java) developers don’t buy in the observability hype
summary
Demo uses a Spring Boot Java app with Micrometer on Google Kubernetes Engine sending metrics to SignalFx and shows concrete pain points: a tiny “hello world” deploy emits ~196 metric types and 2500+ time series, production customers show 900k active series and >200M total, rollups that average five 1s samples hide spikes, and naive trace sampling loses rare slow requests. Builds a Python script that creates dashboards via API to prove observability as code helps; the scripted approach works and prevents accidental dashboard drift.
AI-written summary · about
chapters (14)
- 0:00Introduction and background
- 3:53Definition of observability
- 6:06Demo of a simple microservice
- 8:42Metric explosion in microservices
- 12:21Why observability is needed
- 13:43High cardinality and churn issues
- 17:13Common observability frustrations
- 19:49Confusing tracing charts
- 24:15Alerts and sparse signals
- 26:02Observability as a requirement
- 27:06Auto-instrumentation benefits
- 28:51Manual dashboards problems
- 32:28Observability as code
- 35:02Key takeaways
related talks
01h 10mObservability For (Java) Devs - 2023 Edition
48mAPI Observability by Design: Building Systems That Tell Their Own Stories
53mUnmask hidden issues with Observability superpowers!
52mIs your JVM app flying blind? Unmask hidden issues with Observability superpowers!
51mI Can See Clearly Now: Observability of JVM & Spring Boot 2-3-4 apps
31mObservability for Developers: How to Get There
55mDemystifying Observability for Spring Applications
23mA peek into observability from a tester's lens