Java Logging Framework Benchmark

JMH benchmarks comparing the performance of Java's most popular logging frameworks on Java 25.

Inspired by the SitePoint article on Java logging performance.

Frameworks & Versions

Benchmark Scenarios

Each framework is tested with these scenarios:

Prerequisites

• Java 25 (or later)
• Apache Maven 3.9+

Build

mvn clean package -q

This produces target/benchmarks.jar — a self-contained uber JAR with all dependencies.

Run All Benchmarks

java -jar target/benchmarks.jar

Run Specific Benchmarks

# Only Log4j2 benchmarks
java -jar target/benchmarks.jar "Log4j2"

# Only async benchmarks
java -jar target/benchmarks.jar ".*async.*"

# Only Logback sync
java -jar target/benchmarks.jar "LogbackBenchmark.sync"

# Only JUL
java -jar target/benchmarks.jar "JUL"

Customize JMH Parameters

# Quick smoke test (1 fork, 3 warmup, 5 measurement iterations)
java -jar target/benchmarks.jar -f 1 -wi 3 -i 5

# Output results as JSON
java -jar target/benchmarks.jar -rf json -rff results.json

# List available benchmarks
java -jar target/benchmarks.jar -l

Configuration Details

Log4j2

• Sync: FileAppender with bufferedIO=true, bufferSize=8192
• Async: RandomAccessFileAppender with immediateFlush=false, using Log4j2's AsyncLoggerContextSelector backed by LMAX Disruptor 4.0.0

Logback

• Sync: FileAppender with immediateFlush=true
• Async: AsyncAppender wrapping FileAppender with queueSize=1024, discardingThreshold=0, immediateFlush=false

JUL

• Sync: FileHandler with custom Formatter matching the pattern layout of Log4j2/Logback
• No built-in async support

Common Pattern

All frameworks use an equivalent log pattern:

%date [%thread] %-5level %logger{36} - %message%n

Results Visualization

📊 Live dashboard: brunoborges.github.io/java-logging-benchmark

The dashboard auto-loads the latest CI benchmark results from Linux, macOS, and Windows with tabbed comparison views.

You can also run benchmarks locally and drag & drop your own results.json files onto the page:

java -jar target/benchmarks.jar -rf json -rff results.json

Name files with the OS (e.g. results-linux.json, results-macos.json, results-windows.json) to get per-OS tabs.

CI / Cross-Platform Benchmarks

A GitHub Actions workflow runs the full benchmark suite on three operating systems via a matrix build:

The workflow triggers on pushes and PRs to main, or manually via Actions → Logging Benchmark → Run workflow.

Each runner uploads its results as a JSON artifact (benchmark-results-<os>), making it easy to download and compare cross-platform performance.

Architecture

All benchmarks write to temporary files (via Files.createTempDirectory) to avoid I/O interference between runs. Temp files are cleaned up in @TearDown. Each framework is configured programmatically (Logback, JUL) or via classpath XML config (Log4j2) to ensure isolation.