Application Metrics in Java

Application Metrics in Java | Generated by AI

Home PDF

The com.yammer.metrics library, originally developed by Yammer, is now part of the Dropwizard Metrics library. It provides a robust way to collect and report application metrics in Java, helping you monitor performance aspects like request counts, response times, and more. Below is a guide on how to use it effectively.

Step 1: Add the Dependency

Since com.yammer.metrics has evolved into Dropwizard Metrics, you should use the Dropwizard Metrics library. If you’re using Maven, add the following dependency to your pom.xml:

<dependency>
    <groupId>io.dropwizard.metrics</groupId>
    <artifactId>metrics-core</artifactId>
    <version>4.2.0</version> <!-- Use the latest version available -->
</dependency>

Depending on your needs, you might also want additional modules, such as:

metrics-jvm for JVM-related metrics.
metrics-httpclient for HTTP client metrics.
metrics-jersey for integration with the Jersey web framework.

Check the Dropwizard Metrics documentation for the latest version and available modules.

Step 2: Create a Metric Registry

The MetricRegistry is the central place where all metrics are stored. You typically create one instance for your application:

import com.codahale.metrics.MetricRegistry;

public class MyApplication {
    public static void main(String[] args) {
        MetricRegistry registry = new MetricRegistry();
    }
}

Step 3: Use Different Types of Metrics

Dropwizard Metrics supports several types of metrics, each suited to different monitoring needs:

Counters

Counters are used to track values that can increase or decrease (e.g., number of requests processed).

import com.codahale.metrics.Counter;

Counter counter = registry.counter("requests.processed");
counter.inc();  // Increment by 1
counter.inc(5); // Increment by 5
counter.dec();  // Decrement by 1

Gauges

Gauges provide a snapshot of a value at a specific moment (e.g., current queue size). You define a gauge by implementing the Gauge interface:

import com.codahale.metrics.Gauge;

registry.register("queue.size", new Gauge<Integer>() {
    @Override
    public Integer getValue() {
        return someQueue.size(); // Replace with your logic
    }
});

Histograms

Histograms track the statistical distribution of values (e.g., request sizes):

import com.codahale.metrics.Histogram;

Histogram histogram = registry.histogram("request.sizes");
histogram.update(150); // Record a value

Meters

Meters measure the rate of events (e.g., requests per second):

import com.codahale.metrics.Meter;

Meter meter = registry.meter("requests.rate");
meter.mark(); // Record an event

Timers

Timers measure both the rate and duration of events (e.g., request processing time):

import com.codahale.metrics.Timer;

Timer timer = registry.timer("request.duration");
Timer.Context context = timer.time();
try {
    // Simulate some work
    Thread.sleep(100);
} finally {
    context.stop(); // Record the duration
}

Step 4: Report Metrics

To make metrics useful, you need to report them somewhere. Dropwizard Metrics supports various reporters, such as console, JMX, or Graphite. Here’s an example of a console reporter that logs metrics every 10 seconds:

import com.codahale.metrics.ConsoleReporter;
import java.util.concurrent.TimeUnit;

ConsoleReporter reporter = ConsoleReporter.forRegistry(registry)
    .convertRatesTo(TimeUnit.SECONDS)
    .convertDurationsTo(TimeUnit.MILLISECONDS)
    .build();
reporter.start(10, TimeUnit.SECONDS); // Report every 10 seconds

For production use, consider integrating with systems like Graphite or exposing metrics via JMX.

Step 5: Integrate with Frameworks (Optional)

If you’re using a web framework like Jersey, you can use the metrics-jersey module to automatically instrument your endpoints. Additionally, annotations like @Timed or @Counted can simplify metric collection:

import com.codahale.metrics.annotation.Timed;

@Path("/example")
public class ExampleResource {
    @GET
    @Timed
    public String getExample() {
        return "Hello, world!";
    }
}

Add the metrics-jersey dependency and configure it with your Jersey application to enable this feature.

Best Practices

Naming Conventions: Use a hierarchical naming scheme for clarity (e.g., com.example.myapp.requests.total).
Performance: Be cautious with the number and frequency of metrics, as excessive collection can impact performance.
Documentation: Refer to the Dropwizard Metrics documentation for up-to-date examples and advanced features.

Example: Putting It All Together

Here’s a simple example combining several concepts:

import com.codahale.metrics.*;
import java.util.concurrent.TimeUnit;

public class MetricsExample {
    public static void main(String[] args) throws InterruptedException {
        MetricRegistry registry = new MetricRegistry();

        // Create a counter
        Counter counter = registry.counter("example.counter");
        counter.inc();

        // Create a timer
        Timer timer = registry.timer("example.timer");
        Timer.Context context = timer.time();
        Thread.sleep(100); // Simulate work
        context.stop();

        // Start a console reporter
        ConsoleReporter reporter = ConsoleReporter.forRegistry(registry)
            .convertRatesTo(TimeUnit.SECONDS)
            .convertDurationsTo(TimeUnit.MILLISECONDS)
            .build();
        reporter.start(1, TimeUnit.SECONDS);

        // Keep the program running to see reports
        Thread.sleep(5000);
    }
}

This code tracks a counter and a timer, reporting results to the console every second.

Final Notes

The com.yammer.metrics library has evolved into Dropwizard Metrics, so always use the latest version from the io.dropwizard.metrics group. If you’re starting a new project, you might also explore modern alternatives like Micrometer, but Dropwizard Metrics remains a solid choice for many Java applications. For more details, check the official documentation!

Back 2025.03.06 Donate