The Mechanics of Metrics

on YouTube ↗35m

" summary

Metrics are time‑series key/value pairs; the talk shows concrete pitfalls and fixes by instrumenting a Monster Combat app with Micrometer and Prometheus, then shipping a micrometer extension that merges into Quarkus core. Counters roll over and cause spikes unless converted with the rate() function; tag permutations create a cardinality explosion if uncontrolled. Precomputed client quantiles are unrecombinable across instances, so use histogram buckets or server-side aggregation. The extension and examples prove the approach works in real Quarkus deployments.

AI-written summary · about

" chapters (11)

1. 0:00Introduction
2. 0:50Application Metrics
3. 2:32Time Series
4. 3:20Tags
5. 5:50Counters
6. 11:02Hits and misses
7. 12:02Distribution summaries
8. 13:32Count and sum
9. 21:05Delaying aggregation
10. 25:14Observability
11. 29:56Outro

" related talks

• 36mMetrics - What's your code actually doing?James Burkhart · Open Source Bridge 2013
• 45mMicroProfile Metrics: Practical Use CasesVictor Orozco, Jorge Cajas · Oracle Code One 2018
• 41mRuby Monitoring State of the UnionJoseph Ruscio · RubyConf 2012
• 44mWelcome to the MetricsAlexandre Roman, Didier Burkhalter · SpringOne 2020
• 27mGrand Treatise of Modern Instrumentation and OrchestrationBjörn Rabenstein · Golang UK 2016
• 23mYou Get What You Measure—Why Metrics Are ImportantKumar Srinivasamurthy · SREcon18 Asia/Australia
• 18mJon Schneider at SpringOne Platform 2017Jon Schneider · SpringOne Platform 2017
• 50mHow to Get Started with Metrics?Piotrek Fus · Confitura 2022