State of the Java Metrics Libraries
summary
A small Greeter servlet is instrumented with Prometheus Java client, Micrometer, Dropwizard, MicroProfile and OpenTelemetry and the demos run against Prometheus/Grafana from code on GitHub. Prometheus client is lightweight, exposes 22,000 calls (200) and 4,400 calls (500) in the demo and supports histograms, summaries and exemplars; Micrometer is vendor‑neutral, built into Spring/Quarkus/Micronaut and uses ~12 req/s, ~16% errors from aggregated rates with exponential default buckets; Dropwizard exports precomputed quantiles (no labels), MicroProfile mirrors Dropwizard and is shifting toward Micrometer, OpenTelemetry is promising but alpha.
AI-written summary · about
chapters (18)
- 0:00Introduction
- 2:00Example application
- 6:30Prometheus
- 11:45percentiles
- 15:33Histogram
- 16:00Prometheus Metrics
- 16:43Merging Distributions
- 18:30Recap
- 18:45servlet filter
- 20:08micrometer
- 21:04instrumentation
- 22:50buckets
- 25:12histograms
- 26:40annotations
- 28:37dropwizard metrics
- 30:45microprofile metrics
- 34:18opentelemetry
- 37:01Summary
related talks
45mMicroProfile Metrics: Practical Use Cases
32mOps for Developers - Monitor Your Java Application with Prometheus
38mObservability 101 with Spring and Micrometer
44mDo It Yourself: Programmable Metrics using OpenTelemetry
49mOpenTelemetry for Java Developers
30mPrometheus Monitoring for Java Web Applications w o Modifying Source Code
57mIntroducing Micrometer Application Metrics
53mOpenTelemetry Metrics: observability’s holy grail