Measuring your Elixir Application
summary
Built a metrics dashboard at Xerpa using InfluxDB for storage and Grafana for visualization, with collection from HAProxy, collectd, frontend reports and an Elixir side using exometer/alexer. A Phoenix plug aggregates and reports histograms so 250 ms mean and a 99th percentile near 500 ms show expected uniform-sleep behavior. An example comparison shows one implementation at 0.3 s and the other at 60 s because of a hidden one‑minute sleep, proving measurement beats intuition.
AI-written summary · about
chapters (13)
- 0:00Introduction
- 2:06Why Metrics
- 6:48Time Series
- 9:33Distribution
- 13:36Metrics
- 20:32Demo
- 26:04Metrics Plug
- 31:12Page Controller
- 32:31Shell Script
- 34:27Query Editor
- 36:21Resources
- 37:30Closing remarks
- 39:34Questions
related talks
39mObservable Elixir
39mTelemetry ...and metrics for all
36mMetrics - What's your code actually doing?
37mMonitoring the Health of Your App
34mChallenges in Measuring Erlang Application Performance
25mGreen Benchmarking of BEAM Languages
39mMonitoring Elixir With OpenTelemetry
37mLies, Damn Lies, and Metrics