Java Metrics系统性能监控工具

Metrics是一个Java库，可以对系统进行监控，统计一些系统的性能指标。

比如一个系统后台服务，我们可能需要了解一下下面的一些情况：
1、每秒钟的请求数是多少（TPS）？
2、平均每个请求处理的时间？
3、请求处理的最长耗时？
4、等待处理的请求队列长度？
5、又或者一个缓存服务：缓存的命中率？平均查询缓存的时间？

基本上每一个服务、应用都需要做一个监控系统，这需要尽量以少量的代码，实现统计某类数据的功能。

Metric Registries

MetricRegistry类是Metrics的核心，它是存放应用中所有metrics的容器，也是我们使用 Metrics 库的起点。

MetricRegistry registry = new MetricRegistry();

Metrics 数据展示

Metrics 提供了 Report 接口，用于展示 metrics 获取到的统计数据。metrics-core中主要实现了四种 reporter： JMX ，console， SLF4J，和 CSV。在本文的例子中，我们使用 ConsoleReporter 。

Metrics的五种类型

Gauges

最简单的度量指标，只有一个简单的返回值，例如，我们想衡量一个待处理队列中任务的个数，代码如下：

package com.zyh.maven.metricsdemo;

import com.codahale.metrics.ConsoleReporter;
import com.codahale.metrics.Gauge;
import com.codahale.metrics.MetricRegistry;

import java.util.LinkedList;
import java.util.Queue;
import java.util.concurrent.TimeUnit;

public class GaugeTest {

    public static Queue<String> q = new LinkedList<String>();

    public static void main(String[] args) throws InterruptedException {

        MetricRegistry metricRegistry = new MetricRegistry();
        ConsoleReporter reporter = ConsoleReporter.forRegistry(metricRegistry).build();
        reporter.start(1, TimeUnit.SECONDS);

        metricRegistry.register(MetricRegistry.name(GaugeTest.class, "queue", "size"),
                new Gauge<Integer>(){
                    @Override
                    public Integer getValue() {
                        return q.size();
                    }
                });

        while (true)
        {
            Thread.sleep(1000);
            q.add("张永辉");
        }
    }
}

运行结果

18-2-5 14:36:28 ================================================================

-- Gauges ----------------------------------------------------------------------
com.zyh.maven.metricsdemo.GaugeTest.queue.size
             value = 1


18-2-5 14:36:29 ================================================================

-- Gauges ----------------------------------------------------------------------
com.zyh.maven.metricsdemo.GaugeTest.queue.size
             value = 1


18-2-5 14:36:30 ================================================================

-- Gauges ----------------------------------------------------------------------
com.zyh.maven.metricsdemo.GaugeTest.queue.size
             value = 2


18-2-5 14:36:31 ================================================================

-- Gauges ----------------------------------------------------------------------
com.zyh.maven.metricsdemo.GaugeTest.queue.size
             value = 3

Counters

Counter 就是计数器，Counter 只是用 Gauge 封装了 AtomicLong ，我们可以使用如下的方法获得队列大小，代码如下：

package com.zyh.maven.metricsdemo;

import com.codahale.metrics.ConsoleReporter;
import com.codahale.metrics.Counter;
import com.codahale.metrics.MetricRegistry;

import java.util.Queue;
import java.util.Random;
import java.util.concurrent.LinkedBlockingDeque;
import java.util.concurrent.TimeUnit;

public class CounterTest {

    public static Queue<String> q = new LinkedBlockingDeque<String>();

    public static Counter pendingJobs;

    public static Random random = new Random();

    public static void addJob(String job)
    {
        pendingJobs.inc();
        q.offer(job);
    }

    public static String takeJob()
    {
        pendingJobs.dec();
        return q.poll();
    }

    public static void main(String[] args) throws InterruptedException {

        MetricRegistry registry = new MetricRegistry();
        ConsoleReporter reporter = ConsoleReporter.forRegistry(registry).build();
        reporter.start(1, TimeUnit.SECONDS);

        pendingJobs = registry.counter(MetricRegistry.name(Queue.class, "pending-jobs", "size"));

        int num = 1;
        while(true)
        {
            Thread.sleep(200);
            if(random.nextDouble() > 0.7)
            {
                String job = takeJob();
                System.out.println("take job :" + job);
            }else{
                String job = "Job-" + num;
                addJob(job);
                System.out.println("add Job :" + job);
            }
            num++;
        }
    }
}

运行结果

take job :Job-14
add Job :Job-26
add Job :Job-27
add Job :Job-28
add Job :Job-29
18-2-5 14:39:58 ================================================================

-- Counters --------------------------------------------------------------------
java.util.Queue.pending-jobs.size
             count = 11


take job :Job-16
add Job :Job-31
add Job :Job-32
take job :Job-17
take job :Job-18
18-2-5 14:39:59 ================================================================

-- Counters --------------------------------------------------------------------
java.util.Queue.pending-jobs.size
             count = 10

Meters

Meter度量一系列事件发生的速率(rate)，例如TPS。Meters会统计最近1分钟，5分钟，15分钟，还有全部时间的速率。

package com.zyh.maven.metricsdemo;

import com.codahale.metrics.ConsoleReporter;
import com.codahale.metrics.Meter;
import com.codahale.metrics.MetricRegistry;

import java.util.Random;
import java.util.concurrent.TimeUnit;

public class MeterTest {

    public static Random random = new Random();

    public static void request(Meter meter)
    {
        System.out.println("request");
        meter.mark();
    }

    public static void request(Meter meter, int n)
    {
        while(n > 0)
        {
            request(meter);
            n--;
        }
    }
    public static void main(String[] args) throws InterruptedException {

        MetricRegistry registry = new MetricRegistry();
        ConsoleReporter reporter = ConsoleReporter.forRegistry(registry).build();
        reporter.start(1, TimeUnit.SECONDS);

        Meter meterTps = registry.meter(MetricRegistry.name(MeterTest.class, "request", "tps"));

        while(true)
        {
            request(meterTps, random.nextInt(5));
            Thread.sleep(1000);
        }
    }
}

运行结果

18-2-5 14:42:44 ================================================================

-- Meters ----------------------------------------------------------------------
com.zyh.maven.metricsdemo.MeterTest.request.tps
             count = 16
         mean rate = 2.67 events/second
     1-minute rate = 3.20 events/second
     5-minute rate = 3.20 events/second
    15-minute rate = 3.20 events/second


request
request
request
request
18-2-5 14:42:45 ================================================================

-- Meters ----------------------------------------------------------------------
com.zyh.maven.metricsdemo.MeterTest.request.tps
             count = 20
         mean rate = 2.86 events/second
     1-minute rate = 3.20 events/second
     5-minute rate = 3.20 events/second
    15-minute rate = 3.20 events/second

Histograms

Histogram统计数据的分布情况。比如最小值，最大值，中间值，还有中位数，75百分位，90百分位，95百分位，98百分位，99百分位，和 99.9百分位的值(percentiles)。

package com.zyh.maven.metricsdemo;

import com.codahale.metrics.ConsoleReporter;
import com.codahale.metrics.ExponentiallyDecayingReservoir;
import com.codahale.metrics.Histogram;
import com.codahale.metrics.MetricRegistry;

import java.util.Random;
import java.util.concurrent.TimeUnit;

public class HistogramsTest {

    public static Random random = new Random();

    public static void main(String[] args) throws InterruptedException {

        MetricRegistry registry = new MetricRegistry();
        ConsoleReporter reporter = ConsoleReporter.forRegistry(registry).build();
        reporter.start(1, TimeUnit.SECONDS);

        Histogram histogram = new Histogram(new ExponentiallyDecayingReservoir());
        registry.register(MetricRegistry.name(HistogramsTest.class, "request", "histogram"), histogram);

        while (true)
        {
            Thread.sleep(1000);
            histogram.update(random.nextInt(100000));
        }
    }
}

运行结果

18-2-5 14:45:45 ================================================================

-- Histograms ------------------------------------------------------------------
com.zyh.maven.metricsdemo.HistogramsTest.request.histogram
             count = 8
               min = 8676
               max = 94954
              mean = 36405.28
            stddev = 27543.74
            median = 28243.00
              75% <= 58814.00
              95% <= 94954.00
              98% <= 94954.00
              99% <= 94954.00
            99.9% <= 94954.00


18-2-5 14:45:46 ================================================================

-- Histograms ------------------------------------------------------------------
com.zyh.maven.metricsdemo.HistogramsTest.request.histogram
             count = 9
               min = 8676
               max = 94954
              mean = 39131.65
            stddev = 26922.72
            median = 28243.00
              75% <= 58814.00
              95% <= 94954.00
              98% <= 94954.00
              99% <= 94954.00
            99.9% <= 94954.00

Timers

Timer其实是 Histogram 和 Meter 的结合， histogram 某部分代码/调用的耗时， meter统计TPS。

package com.zyh.maven.metricsdemo;

import com.codahale.metrics.ConsoleReporter;
import com.codahale.metrics.MetricRegistry;
import com.codahale.metrics.Timer;

import java.util.Random;
import java.util.concurrent.TimeUnit;

public class TimerTest {

    public static Random random = new Random();

    public static void main(String[] args) throws InterruptedException {

        MetricRegistry registry = new MetricRegistry();
        ConsoleReporter reporter = ConsoleReporter.forRegistry(registry).build();
        reporter.start(1, TimeUnit.SECONDS);

        Timer timer = registry.timer(MetricRegistry.name(TimerTest.class, "get-latency"));

        Timer.Context ctx;

        while (true)
        {
            ctx = timer.time();
            Thread.sleep(random.nextInt(1000));
            ctx.stop();
        }
    }
}

运行结果

18-2-5 14:48:30 ================================================================

-- Timers ----------------------------------------------------------------------
com.zyh.maven.metricsdemo.TimerTest.get-latency
             count = 30
         mean rate = 2.15 calls/second
     1-minute rate = 2.02 calls/second
     5-minute rate = 2.00 calls/second
    15-minute rate = 2.00 calls/second
               min = 22.82 milliseconds
               max = 987.23 milliseconds
              mean = 439.66 milliseconds
            stddev = 263.14 milliseconds
            median = 421.99 milliseconds
              75% <= 582.73 milliseconds
              95% <= 926.66 milliseconds
              98% <= 987.23 milliseconds
              99% <= 987.23 milliseconds
            99.9% <= 987.23 milliseconds

上面写了几个demo尝试用了一下Metrics，在这里记录一下！