This library provides a simple way to report metrics for Java applications based on the Dropwizard Metrics Project and the Async Properties library.
Why Other metric frameworks are often complex and feature-rich. This framework is composed of a small set of core classes; thus, it is easy to understand and plug into various environments.
For instance: How many users are logged into the system concurrently?
PropertyNode m = Metrics.create();
m.record(Metrics.increment("users"));
m.record(Metrics.increment("users"));
m.record(Metrics.decrement("users"));
m.print();
m.stop().get();
Should result in the output:
{
'users': '1'
}
For instance: How many times per second is a service called?
final PropertyNode m = Metrics.create();
m.record(Metrics.happened("serviceCalled"));
m.record(Metrics.happened("serviceCalled"));
m.record(Metrics.happened("serviceCalled"));
Thread.sleep(10000);
m.print();
m.stop().get();
Should result in the output:
{
'serviceCalled': {
'Total Events': '3',
'Events per Second (last Minute)': '0.552026648777594',
'Events per Second (last 5 Minutes)': '0.5900828722929705',
'Events per Second (last 15 Minutes)': '0.5966759088029381'
}
}
For instance: How large are received requests in average?
final PropertyNode m = Metrics.create();
m.record(Metrics.value("requestSize", 300));
m.record(Metrics.value("requestSize", 100));
m.record(Metrics.value("requestSize", 200));
m.print();
m.stop().get();
Should result in.
{
'requestSize': {
'Mean': '200.0',
'Standard Deviation': '100.0',
'Max': '300',
'Min': '100',
'75% of Values Smaller Than': '300.0',
'95% of Values Smaller Than': '300.0',
'98% of Values Smaller Than': '300.0'
}
}
Note: Statistics are by default calculated using so called Exponentially Decaying Reservoirs. These essentially will provide the values and averages for the last 5 minutes of application runtime.
Metrics are turned into easily human-readable strings. What you do with these is up to you.
PropertyNode m = ...
String metric = m.retrieve("metricId").get().toString()
PropertyNode m = ...
String metrics = m.render().get();
Just add the following dependency to your projects.
<dependency>
<groupId>org.javadelight</groupId>
<artifactId>delight-metrics</artifactId>
<version>[insert latest version]</version>
</dependency>
This artifact is available on Maven Central and BinTray.
This project is compatible with the following environments:
- Java 1.6+
- GWT 2.5.0+
- Android (any)
- OSGi (any)
A significant part of the internal implementation in the library is based on the Dropwizard Metrics Project. Thus, some of the performance characteristics are inherited from this project.
However, this project uses a different approach to multi-threading. While in the Dropwizard Metrics Project every metric and operation is synchronized independently, the Lightweight Java Metrics project synchronizes all operations in one place (see SynchronizedMetricsNode.java).
Thus, more operations can be performed within the scope of one thread. This might lead to better performance as compared to the Dropwizard Metrics Project in some case and to worse performance in others.