Package io.prometheus.client

Class Histogram

java.lang.Object
io.prometheus.client.Collector
io.prometheus.client.SimpleCollector<Histogram.Child>
io.prometheus.client.Histogram
All Implemented Interfaces:
Collector.Describable
public class Histogram extends SimpleCollector<Histogram.Child> implements Collector.Describable
Histogram metric, to track distributions of events.

Example of uses for Histograms include:

  • Response latency
  • Request size

Note: Each bucket is one timeseries. Many buckets and/or many dimensions with labels can produce large amount of time series, that may cause performance problems.

The default buckets are intended to cover a typical web/rpc request from milliseconds to seconds.

Example Histograms: 

 
   class YourClass {
     static final Histogram requestLatency = Histogram.build()
         .name("requests_latency_seconds").help("Request latency in seconds.").register();

     void processRequest(Request req) {
        Histogram.Timer requestTimer = requestLatency.startTimer();
        try {
          // Your code here.
        } finally {
          requestTimer.observeDuration();
        }
     }

     // Or if using Java 8 lambdas.
     void processRequestLambda(Request req) {
        requestLatency.time(() -> {
          // Your code here.
        });
     }
   }

You can choose your own buckets: 

 
     static final Histogram requestLatency = Histogram.build()
         .buckets(.01, .02, .03, .04)
         .name("requests_latency_seconds").help("Request latency in seconds.").register();
linearBuckets and exponentialBuckets offer easy ways to set common bucket patterns.