ContiPerf接口性能测试

13 开源组件 评论

1. 简介

ContiPerf是一个轻量级的测试工具，基于JUnit4开发，可用于接口级的性能测试，可以设置线程数和执行次数，通过限制最大时间和平均执行时间来进行效率测试。

2. 使用方法

  1. 添加依赖

    1
    2
    3
    4
    5
    6
    <dependency>
        <groupId>org.databene</groupId>
        <artifactId>contiperf</artifactId>
        <version>2.3.4</version>
        <scope>test</scope>
    </dependency>

  2. 创建Domain

    1
    2
    3
    4
    5
    6
    7
    8
    9
    10
    11
    12
    13
    14
    15
    16
    17
    18
    @Entity
    @Table(name = "g_data_sources")
    @org.hibernate.annotations.Table(appliesTo = "g_data_sources", comment = "数据源信息表")
    public class DataSources extends BasePO {

        @Column(name = "name", columnDefinition = "varchar(30) NOT NULL comment '数据库名'")
        private String name;

        @Column(name = "url", columnDefinition = "varchar(100) NOT NULL comment '数据库ip'")
        private String url;

        @Column(name = "username", columnDefinition = "varchar(30) NOT NULL comment '数据库连接用户名'")
        private String username;

        @Column(name = "pwd", columnDefinition = "varchar(30) default '' comment '数据库连接密码'")
        private String pwd;
        //-----------------省略getter/setter方法 ---------------
    }

  3. 创建查询代码

    1
    2
    3
    4
    5
    6
    7
    8
    9
    10
    11
    12
    13
    14
    15
    @Repository("dataSourcesDaoImpl")
    @Transactional(propagation = Propagation.REQUIRED, rollbackFor = Exception.class)
    public class DataSourcesDaoImpl {

        @PersistenceContext
        private EntityManager entityManager;
        
        public DataSources findById(Integer id) {
            List<DataSources> result = entityManager.createQuery("from DataSources e where e.id = " + id).getResultList();
            if (result != null && !result.isEmpty()) {
                return result.get(0);
            }
            return null;
        }
    }

  4. 创建测试类

    1
    2
    3
    4
    5
    6
    7
    8
    9
    10
    11
    12
    13
    14
    15
    16
    17
    18
    19
    20
    21
    @Slf4j
    @SpringBootTest
    @RunWith(SpringRunner.class)
    public class MybatisGeneratorApplicationTest {

        @Autowired
        private DataSourcesDaoImpl dataSourcesDao;

        @Rule
        public ContiPerfRule rule = new ContiPerfRule();    //重点

        @Test
        //10个线程 执行100次
        @PerfTest(invocations = 100, threads = 10)           //重点
        //指定每次执行的最长时间/平均时间/总时间
        // @Required(max = 1200, average = 250, totalTime = 60000)
        public void test() {
            int id = (int) (Math.random() * 60);
            dataSourcesDao.findById(id);
        }
    }

  5. JUnit执行完毕，会在target/contiperf-report中有相关的执行结果

  1. 可以使用浏览器打开查看结果

    • Measured invocations: 请求次数
    • Thread Count: 线程数
    • Execution time：总执行时间
    • Throughput: 吞吐量，每秒效率 TPS
    • Min. latency: 最短响应时间
    • Average latency: 平均响应时间
    • Median: TP50响应时间
    • 90%: TP90响应时间，指在一个时间段内（如5分钟），统计该方法每次调用所消耗的时间，并将这些时间按从小到大的顺序进行排序，取第90%的那个值作为TP90 值；配置此监控指标对应的报警阀值后，需要保证在这个时间段内该方法所有调用的消耗时间至少有90%的值要小于此阀值，否则系统将会报警
    • Max latency: 最长响应时间

报告图片显示不出来，是源码中使用了google图表，需要在线！！并且源码中写死了cht=lxy，可以参考ContiPerf html报告

3. @PerfTest参数说明

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
@Documented
@Target({ElementType.METHOD, ElementType.TYPE})
@Retention(RetentionPolicy.RUNTIME)
public @interface PerfTest {
    //常用的就是这个参数 执行次数  与线程无关
    int invocations() default 1;
   /**
    * 间隔时间  可以暂时不用  因为性能测试主要是测试并发
    * @PerfTest(invocations = 300, threads = 2, duration = 100)，如果执行方法300次的时候执行时间还没到100ms，则继续执行到满足执行时间等于100ms，如果执行到50次的时候已经100ms了，则会继续执行之100次。
    * The number of milliseconds to run and repeat the test with the full number of configured threads -
    * use this alternatively to {@link #invocations()}. When using a {@link #rampUp()}, the ramp-up times
    * add to the duration.
    * @see #duration()
    */
    int duration() default -1;
   /**
    * 线程数 
    * The number of threads which concurrently invoke the test. The default value is 1.
    */
    int threads() default 1;
   /**
    * The number of milliseconds to wait before each thread is added to the currently active threads.
    * On {@link #duration()}-based tests, the total ramp-up time of rampUp * (threads - 1) is added to the
    * configured duration.
    */
    int rampUp() default 0;
    
   /** 
    * The number of milliseconds to wait before the actual measurement and requirements monitoring is activated.
    *  Use this to exclude ramp-up times from measurement or wait some minutes before dynamic optimizations are
    *  applied (like code optimization or cache population). 
    */
    int warmUp() default 0;
   /** 
    * Set this to true, if execution should stop with a failure message as soon as a configured {@link Required#max()}
    * value is violated. Set it to false, if you are interested in performing a full measurement to get percentiles,
    * throughput and more. The default value is false. 
    */
    boolean cancelOnViolation() default false;
    
   /** 
    * The class of a {@link WaitTimer} implementation by which a wait time can be incurred between test invocations 
    */
    Class<? extends WaitTimer> timer() default None.class;
   /** 
    * The parameters to initialize the {@link WaitTimer}.
    * The meaning of the values is individual for the WaitTimer implementation. 
    */
    double[] timerParams() default {};
   /** 
    * One ore more {@link Clock} classes to use for time measurement.
    * The first one specified is the one relevant for requirements verification. 
    */
    Class<? extends Clock>[] clocks() default {};
}

4. @Required参数说明

参数 说明
@Required(throughput = 20) 要求每秒至少执行20个测试
@Required(average = 50) 要求平均执行时间不超过50ms
@Required(median = 45) 要求所有执行的50%不超过45ms
@Required(max = 2000) 要求没有测试超过2s
@Required(totalTime = 5000) 要求总的执行时间不超过5s
@Required(percentile90 = 3000) 要求90%的测试不超过3s
@Required(percentile95 = 5000) 要求95%的测试不超过5s
@Required(percentile99 = 10000) 要求99%的测试不超过10s

小余哥xiaoyouge

分享一些技术教程与实践，以及日常踩坑记录。