0Pricing
GraphQL APIs with Spring Boot · レッスン

バッチ処理とキャッシュの実装

DataLoadersをSpring BootのResolverに統合し、データ取得を最適化してデータベースへの往復回数を減らします。

「バッチ処理とキャッシュの実装」はCoddyKit上の無料GraphQL APIs with Spring Bootレッスンです。 これはレッスン3/4です。 下記で完全なレッスンを無料で読むことができます。その後、ブラウザ内の組み込みコードエディタと24時間対応のAIチューターでハンズオン演習できます。 これはGraphQL APIs with Spring Boot学習パスの一部であり、ウェブとCoddyKitアプリ全体で進捗が同期されます。 GraphQL APIs with Spring Bootコースには全4レッスンが含まれています。

このレッスンの一部はまだ翻訳されておらず、英語で表示されています。

DataLoaders: Time to Implement!

Welcome back! In previous lessons, we learned about the N+1 problem and how DataLoaders provide an elegant solution through batching and caching.

Today, we'll get hands-on and integrate DataLoaders into a Spring Boot GraphQL application. We'll see how to define a BatchLoader, register it, and use it in our GraphQL resolvers.

Simulating a Data Service

First, let's set up a simple mock service that simulates fetching users from a database. This service will be called by our DataLoaders.

Notice the System.out.println, which will help us observe when the actual 'database' call happens, demonstrating batching later.

import java.util.List;
import java.util.Map;
import java.util.stream.Collectors;

// A simple User data class
class User {
  String id;
  String name;
  User(String id, String name) { 
    this.id = id; this.name = name; 
  }
  public String getId() { return id; }
  public String getName() { return name; }
}

// Mock service to fetch users (simulates DB call)
class MockUserService {
  private final Map<String, User> users = Map.of(
    "1", new User("1", "Alice"),
    "2", new User("2", "Bob"),
    "3", new User("3", "Charlie"),
    "4", new User("4", "David")
  );

  public List<User> findAllByIds(List<String> ids) {
    System.out.println("DB Call: Fetching users for IDs: " + ids);
    return ids.stream()
              .map(users::get)
              .filter(user -> user != null)
              .collect(Collectors.toList());
  }
}

public class Main {
  public static void main(String[] args) {
    MockUserService service = new MockUserService();
    List<User> foundUsers = service.findAllByIds(List.of("1", "3"));
    System.out.println("Found: " + foundUsers.size() + " users.");
  }
}

The BatchLoader Interface

The core of DataLoader's batching mechanism is the BatchLoader interface. It defines a single method: load(List<K> keys).

  • It takes a list of keys (e.g., user IDs).
  • It returns a CompletionStage (a future-like object) of a list of values (e.g., users).
  • Crucially, this method is called once for all keys requested in a short window, allowing you to fetch them in a single optimized operation (like a single database query).

Implementing Your BatchLoader

Now, let's create our UserBatchLoader. It will implement the BatchLoader<String, User> interface, meaning it takes a String (user ID) as a key and returns a User object.

It uses our MockUserService to perform the actual batch fetch.

import org.dataloader.BatchLoader;
import java.util.List;
import java.util.concurrent.CompletableFuture;
import java.util.concurrent.CompletionStage;

// (Assume User and MockUserService are defined elsewhere)
// For runnable example, they would be in the same file.

public class UserBatchLoader implements BatchLoader<String, User> {
  private final MockUserService userService;

  public UserBatchLoader(MockUserService userService) {
    this.userService = userService;
  }

  @Override
  public CompletionStage<List<User>> load(List<String> userIds) {
    // This method is called once with all requested IDs.
    // We delegate to our mock service to fetch them in one go.
    return CompletableFuture.supplyAsync(
      () -> userService.findAllByIds(userIds)
    );
  }
}

public class Main {
  public static void main(String[] args) {
    MockUserService userService = new MockUserService();
    UserBatchLoader batchLoader = new UserBatchLoader(userService);
    System.out.println("UserBatchLoader created!");
  }
}

The DataLoaderRegistry

In a Spring Boot GraphQL application, DataLoader instances are managed within a DataLoaderRegistry. This registry acts as a container for all DataLoaders available during a GraphQL request execution.

  • Each DataLoader is registered with a unique name (e.g., "userDataLoader").
  • Resolvers retrieve the necessary DataLoader from this registry using its name.
  • Spring for GraphQL typically handles the lifecycle and scope of this registry per request.

Setting up DataLoaders

Here's how you create a DataLoader instance from your BatchLoader and register it in a DataLoaderRegistry. This setup is crucial for making DataLoaders accessible to your resolvers.

import org.dataloader.DataLoader;
import org.dataloader.DataLoaderRegistry;

// (Assume User, MockUserService, UserBatchLoader are defined)

public class Main {
  public static void main(String[] args) {
    MockUserService userService = new MockUserService();
    UserBatchLoader userBatchLoader = new UserBatchLoader(userService);

    // Create a DataLoader instance
    DataLoader<String, User> userDataLoader =
      DataLoader.newDataLoader(userBatchLoader);

    // Register it in a DataLoaderRegistry
    DataLoaderRegistry registry = new DataLoaderRegistry();
    registry.register("userDataLoader", userDataLoader);

    System.out.println("DataLoader registered as 'userDataLoader'!");
    // In a real app, this registry would be part of the GraphQLContext.
    // Resolvers would then retrieve DataLoaders from it.
  }
}

Using DataLoader in Resolvers

Once your DataLoader is registered, you can use it within your GraphQL resolvers. Instead of directly calling your service, you'll call dataLoader.load(id).

  • The load() method returns a CompletionStage, not the direct object.
  • The DataLoader collects all load() calls within the current execution frame before dispatching them to the BatchLoader.
  • Spring for GraphQL automatically dispatches the DataLoaders at the end of a data fetching cycle.

Resolver Integration in Action

This example simulates a GraphQL query where multiple user IDs are requested. Watch the "DB Call" message in the output to see how DataLoaders batch these requests into a single operation, even if they appear separate in the resolver logic.

import org.dataloader.DataLoader;
import org.dataloader.DataLoaderRegistry;
import java.util.List;
import java.util.Map;
import java.util.concurrent.CompletableFuture;
import java.util.concurrent.CompletionStage;
import java.util.stream.Collectors;

// User, MockUserService, UserBatchLoader definitions (from previous scenes)
class User {
  String id; String name;
  User(String id, String name) { this.id = id; this.name = name; }
  public String getId() { return id; }
  public String getName() { return name; }
}

class MockUserService {
  private final Map<String, User> users = Map.of(
    "1", new User("1", "Alice"), "2", new User("2", "Bob"),
    "3", new User("3", "Charlie"), "4", new User("4", "David")
  );
  public List<User> findAllByIds(List<String> ids) {
    System.out.println("DB Call: Fetching users for IDs: " + ids);
    return ids.stream().map(users::get)
              .filter(user -> user != null).collect(Collectors.toList());
  }
}

class UserBatchLoader implements BatchLoader<String, User> {
  private final MockUserService userService;
  public UserBatchLoader(MockUserService userService) { this.userService = userService; }
  @Override
  public CompletionStage<List<User>> load(List<String> userIds) {
    return CompletableFuture.supplyAsync(() -> userService.findAllByIds(userIds));
  }
}

// Simulate a GraphQL Query Resolver method
public class UserResolver {
  private final DataLoaderRegistry dataLoaderRegistry;

  public UserResolver(DataLoaderRegistry registry) {
    this.dataLoaderRegistry = registry;
  }

  public CompletionStage<User> getUserById(String id) {
    DataLoader<String, User> userDataLoader =
      dataLoaderRegistry.getDataLoader("userDataLoader");
    return userDataLoader.load(id);
  }

  public static void main(String[] args) {
    MockUserService userService = new MockUserService();
    UserBatchLoader userBatchLoader = new UserBatchLoader(userService);
    DataLoader<String, User> userDataLoader =
      DataLoader.newDataLoader(userBatchLoader);

    DataLoaderRegistry registry = new DataLoaderRegistry();
    registry.register("userDataLoader", userDataLoader);

    UserResolver resolver = new UserResolver(registry);

    System.out.println("Requesting User 1 and User 2 concurrently...");

    // Simulate two concurrent GraphQL field fetches
    CompletableFuture<User> user1Future =
      (CompletableFuture<User>) resolver.getUserById("1");
    CompletableFuture<User> user2Future =
      (CompletableFuture<User>) resolver.getUserById("2");

    CompletableFuture.allOf(user1Future, user2Future)
      .thenRun(() -> {
        try {
          System.out.println("Fetched User 1: " + user1Future.join().getName());
          System.out.println("Fetched User 2: " + user2Future.join().getName());
        } catch (Exception e) {
          e.printStackTrace();
        }
        // For this manual demo, we dispatch the DataLoader.
        // In Spring GraphQL, this is handled automatically.
        userDataLoader.dispatch();
      }).join();
  }
}

DataLoader's Built-in Caching

Beyond batching, DataLoaders also provide request-scoped caching out-of-the-box. This means:

  • If dataLoader.load("1") is called multiple times within the same GraphQL request, the BatchLoader's fetch function for ID "1" will only be invoked once.
  • Subsequent calls for the same ID within that request will return the already fetched (cached) result.
  • This prevents redundant database calls for the same data within a single GraphQL operation, further boosting performance.

Check Your Understanding

You've learned how to implement DataLoaders for batching and caching. Let's test your knowledge!

Recap: Batching & Caching

Great job! You've successfully learned how to implement DataLoaders in Spring Boot GraphQL.

  • We created a MockUserService to simulate data fetching.
  • We built a UserBatchLoader to handle batching multiple ID requests.
  • We saw how to initialize and register DataLoader instances in a DataLoaderRegistry.
  • Finally, we integrated DataLoaders into a resolver, observing how requests are batched and how caching prevents redundant calls.

By using DataLoaders, you significantly optimize your GraphQL API's performance, especially when dealing with complex data graphs and nested objects.

よくある質問

「バッチ処理とキャッシュの実装」レッスンは無料ですか？

はい。「バッチ処理とキャッシュの実装」の完全なテキストはこのウェブで無料で読めます。インタラクティブに演習し（組み込みコードエディタと24時間対応のAIチューター）、GraphQL APIs with Spring Bootコースの残りをアンロックするには、CoddyKit PROにアップグレードしてください。 GraphQL APIs with Spring Bootコースには全4レッスンが含まれています。

「バッチ処理とキャッシュの実装」で何を学びますか？

DataLoadersをSpring BootのResolverに統合し、データ取得を最適化してデータベースへの往復回数を減らします。 ブラウザで直接実行するハンズオンコードでGraphQL APIs with Spring Bootを演習し、24時間対応のAIチューターがレッスンを進める中での質問に答えます。

GraphQL APIs with Spring Bootを始めるのに経験は必要ですか？

事前経験は必要ありません。CoddyKitのGraphQL APIs with Spring Bootは初級者から上級者向けに構成されているため、ここから始めるか最初から始めて、自分のペースで進むことができます。 これはレッスン3/4です。

「バッチ処理とキャッシュの実装」レッスンにはどのくらい時間がかかりますか？

ほとんどのCoddyKitレッスンは約5～10分かかります。各レッスンはコンパクトでインタラクティブなので、着実に進歩し、ウェブとアプリ全体で正確に前回の場所から再開できます。

このGraphQL APIs with Spring Bootレッスンでコードを書いて実行できますか？

はい。すべてのGraphQL APIs with Spring Bootレッスンに組み込みコードエディタが含まれているため、ブラウザでリアルコードを書いて実行し、即座のAIフィードバックを取得できます。ローカル設定は不要です。

このコースのすべてのレッスン

  1. N+1問題を解説
  2. GraphQL DataLoadersの紹介
  3. バッチ処理とキャッシュの実装
  4. Springコンテキストと非同期処理で使うDataLoader
← GraphQL APIs with Spring Bootに戻る