实战示例

1. 添加maven依赖：

<parent>
    <groupId>org.springframework.boot</groupId>
    <artifactId>spring-boot-starter-parent</artifactId>
    <version>2.0.4.RELEASE</version>
</parent>

    <dependencies>
        <dependency>
            <groupId>org.springframework.boot</groupId>
            <artifactId>spring-boot-starter-webflux</artifactId>
        </dependency>
        <dependency>
            <groupId>org.projectlombok</groupId>
            <artifactId>lombok</artifactId>
        </dependency>
    </dependencies>

    <build>
        <plugins>
            <plugin>
                <groupId>org.springframework.boot</groupId>
                <artifactId>spring-boot-maven-plugin</artifactId>
                <executions>
                    <execution>
                        <goals>
                            <goal>repackage</goal>
                        </goals>
                    </execution>
                </executions>
            </plugin>
        </plugins>
    </build>

2. 添加启动类

@SpringBootApplication
public class RetailApp {
  
  public static void main(String[] args){
      SpringApplication.run(RetailApp.class, args);
  }
}

3. 添加领域实体和服务类

@Data
@Accessors(chain = true)
public class Spu {
  private Long id;
  private String name;
}
@Service
public class SpuService {
  
  public Spu findSpuById(Long id){
    try {
      TimeUnit.MILLISECONDS.sleep(50); //模拟耗时任务
    } catch (InterruptedException e) {
      e.printStackTrace();
    }
    return new Spu().setId(id).setName("spu"+id);
  }
}

4. 添加rest api

@RestController
@Slf4j
public class SpuRestApi {
  
  @Autowired
  SpuService spuService;
  
  @GetMapping("/v1/spu/detail")
  public Mono<Spu> findById(Long id){
    StopWatch stopWatch = new StopWatch();
    stopWatch.start();
    return Mono.just(id)
        .flatMap(x-> Mono.fromCallable(()-> spuService.findSpuById(x))
        .subscribeOn(Schedulers.elastic())) //关键，netty线程不适合用来处理长时间阻塞的任务
        .doOnSuccess(x-> {
          stopWatch.stop();
          log.info("{}", stopWatch.toString());
        });
  }
}

压力测试

压测环境：

• macbook pro
• 2.7 GHz Intel Core i7
• 16 GB

从压测结果来看，可以达到惊人的1.5w+以上的QPS，系统表现还相对稳定。

wrk -t10 -c100 -d120s --timeout 2s --latency 'http://localhost:8080/v1/spu/detail?id=1'
Running 2m test @ http://localhost:8080/v1/spu/detail?id=1
  10 threads and 100 connections
  Thread Stats   Avg      Stdev     Max   +/- Stdev
    Latency    54.09ms    2.71ms 163.71ms   86.00%
    Req/Sec   185.59     17.25   203.00     85.88%
  Latency Distribution
     50%   54.29ms
     75%   55.26ms
     90%   55.94ms
     99%   58.27ms
  221895 requests in 2.00m, 22.64MB read
Requests/sec:   1847.68
Transfer/sec:    193.07KB


wrk -t10 -c500 -d120s --timeout 2s --latency 'http://localhost:8080/v1/spu/detail?id=1'
Running 2m test @ http://localhost:8080/v1/spu/detail?id=1
  10 threads and 500 connections
  Thread Stats   Avg      Stdev     Max   +/- Stdev
    Latency    55.44ms    6.89ms 294.20ms   97.99%
    Req/Sec     0.89k   108.62     1.01k    91.97%
  Latency Distribution
     50%   55.14ms
     75%   56.32ms
     90%   57.70ms
     99%   66.94ms
  1067897 requests in 2.00m, 108.97MB read
  Socket errors: connect 0, read 464, write 0, timeout 0
Requests/sec:   8894.89
Transfer/sec:      0.91MB


wrk -t10 -c1000 -d120s --timeout 2s --latency 'http://localhost:8080/v1/spu/detail?id=1'
Running 2m test @ http://localhost:8080/v1/spu/detail?id=1
  10 threads and 1000 connections
  Thread Stats   Avg      Stdev     Max   +/- Stdev
    Latency    61.92ms   20.97ms 536.55ms   94.69%
    Req/Sec     1.58k   294.84     2.02k    83.12%
  Latency Distribution
     50%   57.85ms
     75%   62.72ms
     90%   73.75ms
     99%  118.44ms
  1875346 requests in 2.00m, 191.37MB read
  Socket errors: connect 0, read 1081, write 0, timeout 0
Requests/sec:  15616.09
Transfer/sec:      1.59MB

用VisualVM跟踪如下：

压测跟踪.png

spring-boot webflux配置图

webflux config.png

webflux核心组件示意图

webflux component.png