1. Start mockserver
   1. Go to mockserver folder and type docker-compose up -d
   2. Go to http://localhost:1080/mockserver/dashboard to see loaded expectations
   3. No mockserver restart needed when updating cfdata.json
2. Start spring boot app CfApplication: ./mvnw spring-boot:run
3. Go to http://localhost:8080/productsv1
4. Look at log to check duration of the call

Scenario:

Return list of products including price, stock level, reviews with 5 different services

1. http://localhost:1080/products list products including productId
2. http://localhost:1080/prices?productId={} price for product (May not produce error or no data, if error, skip product)
3. http://localhost:1080/warehouses?productId={} returns locationId for stock query (if error return stock unknown)
4. http://localhost:1080/locations?locationId={}&productId={} returns stock for product (if error return stock unknown)
5. http://localhost:1080/reviews?productId={} list of reviews, on error empty list

ProductDataService Blocking calls for all

Steps:

1. Open ProductService2 and analyze each call
2. Convert calls into CompletableFuture calls
3. Update code to run futures in parallel where applicable
4. Go to http://localhost:8080/productsv2
5. Compare duration in logs with duration of productsv1 call
6. Make it even faster!!

Starting tips

1. Changing call to ComputableFuture

private CompletableFuture<RemoteProductPrices> getProductPrice(String productId) {
   return CompletableFuture.supplyAsync(() -> {
       return restTemplate.getForObject("/prices?productId=" + productId, RemoteProductPrices.class);
   });
}

2. Handling exceptions .exceptionally

3. Use result of ComputalbleFuture or create something new .thenApply "map" or .thenCompose "flatMap"

4. User custom thread pool for execution, tweak values for more performance

    ThreadPoolTaskExecutor myExcutor = new ThreadPoolTaskExecutor();
    myExcutor.setCorePoolSize(2);
    myExcutor.setMaxPoolSize(2);
    myExcutor.setQueueCapacity(500);
    myExcutor.setThreadNamePrefix("MyCustomPool-");
    myExcutor.initialize();

more info:

https://howtodoinjava.com/java/multi-threading/java-thread-pool-executor-example/

https://dzone.com/articles/be-aware-of-forkjoinpoolcommonpool

5. Monitor threads by looking at spring boot metrics actuator. Add management.endpoints.web.exposure.include: '*' to application.yml and create metric for your pool as in example below

  @Bean
  public ExecutorService myExecutor(final MeterRegistry registry) {
    return ExecutorServiceMetrics
        .monitor(registry, Executors.newFixedThreadPool(10), "myExecutorPool");
  }

6. Configure HttpClient for your rest templates

  @Bean
  public HttpClient myHttpClient(MyProps properties) {
    HttpClientBuilder httpClientBuilder = HttpClientBuilder.create()
        .setDefaultRequestConfig(
            RequestConfig.custom()
                .setConnectTimeout(Math.toIntExact(properties.getConnectionTimeout().toMillis()))
                .setSocketTimeout(Math.toIntExact(properties.getReadTimeout().toMillis()))
                .build())
        .setMaxConnTotal(100)
        .setMaxConnPerRoute(50);
    return httpClientBuilder.build();
  }

  @Bean
  RestTemplate myTemplate(RestTemplateBuilder builder, MyProps properties) {
    return builder
        .requestFactory(() -> new HttpComponentsClientHttpRequestFactory(myHttpClient(properties)))
        .rootUri(properties.getRootUri())
        .build();
  }

7. Wait for many to complete

CompletableFuture.allOf(futures.toArray(new CompletableFuture[futures.size()]))