GraphQL DataLoader Pattern

GraphQL DataLoader Pattern

Batch and cache data fetches to eliminate N+1 queries in GraphQL resolvers

When to Use

• Field resolvers execute per item in a list, causing N+1 database queries
• Multiple resolvers in the same request need the same record
• You need request-scoped caching without a shared application cache
• Optimizing GraphQL API response times without restructuring the schema

Instructions

1. Create one DataLoader instance per request, per resource type. DataLoader batches all .load(key) calls within a single tick of the event loop into one batch function call. It must be request-scoped to avoid leaking cached data between users.

import DataLoader from 'dataloader';

function createLoaders(db: Database) {
  return {
    userById: new DataLoader<string, User>(async (ids) => {
      const users = await db.users.findByIds([...ids]);
      const userMap = new Map(users.map((u) => [u.id, u]));
      return ids.map((id) => userMap.get(id) || new Error(`User ${id} not found`));
    }),
  };
}

2. The batch function must return results in the same order as the input keys. This is the critical contract. If key index 0 is "abc", result index 0 must be the value for "abc" or an Error. Use a Map to reorder database results to match the input key order.

3. Attach loaders to the GraphQL context. Create fresh loaders for each request in your server's context factory.

const server = new ApolloServer({
  typeDefs,
  resolvers,
  context: ({ req }) => ({
    currentUser: authenticate(req),
    loaders: createLoaders(db),
  }),
});

4. Use .load() in field resolvers instead of direct database calls.

const resolvers = {
  Order: {
    customer: (order, _args, { loaders }) => {
      return loaders.userById.load(order.customerId);
    },
  },
  Comment: {
    author: (comment, _args, { loaders }) => {
      return loaders.userById.load(comment.authorId);
    },
  },
};

When a query returns 50 orders, the customer field resolver calls .load() 50 times, but DataLoader batches them into a single findByIds([...50 ids]) call.

5. Use .loadMany() for one-to-many relationships where you have an array of keys.

const resolvers = {
  User: {
    favoriteProducts: (user, _args, { loaders }) => {
      return loaders.productById.loadMany(user.favoriteProductIds);
    },
  },
};

6. Create loaders for different access patterns. A userById loader and a userByEmail loader are separate DataLoader instances with separate batch functions and separate caches.

7. Use .prime() to warm the cache when you already have the data. After a mutation creates a user, prime the loader so subsequent reads in the same request hit the cache.

const newUser = await db.users.create(input);
context.loaders.userById.prime(newUser.id, newUser);

8. Disable caching for loaders where stale reads are dangerous. Pass { cache: false } to the DataLoader constructor if you need batching without memoization — rare, but useful for data that changes mid-request.

Details

How batching works: DataLoader collects all .load() calls made synchronously (within the same microtask/tick). At the end of the tick, it calls the batch function once with all collected keys, then distributes results back to each caller. This turns N sequential queries into 1 batched query.

Per-request scope is mandatory. If you reuse a DataLoader across requests, User A's data will be served from cache to User B. Always create new DataLoader instances in the context factory.

Batch function constraints:

• Must accept an array of keys and return a Promise of an array of values
• Return array length must equal input keys length
• Return array order must match input keys order
• Individual errors are represented as Error instances at the corresponding index

Nested batching: DataLoader batches within a single tick, but GraphQL resolves level by level. This means field resolvers at depth 2 are batched separately from depth 3. This is usually fine — each level gets one batch call per loader.

Composite keys: When the lookup requires multiple values (e.g., orgId + userId), serialize the key to a string and provide a cacheKeyFn:

new DataLoader<{ orgId: string; userId: string }, Membership>(
  async (keys) => {
    /* batch by org+user */
  },
  { cacheKeyFn: (key) => `${key.orgId}:${key.userId}` }
);

Common mistakes:

• Creating DataLoader outside request scope (data leaks between users)
• Returning results in database order instead of input key order (mismatched results)
• Using DataLoader for write operations (it is a read optimization)
• Not handling missing records (returning undefined instead of Error breaks the contract)

Source

https://github.com/graphql/dataloader

Process

1. Read the instructions and examples in this document.
2. Apply the patterns to your implementation, adapting to your specific context.
3. Verify your implementation against the details and edge cases listed above.

Harness Integration

• Type: knowledge — this skill is a reference document, not a procedural workflow.
• No tools or state — consumed as context by other skills and agents.

Success Criteria

• The patterns described in this document are applied correctly in the implementation.
• Edge cases and anti-patterns listed in this document are avoided.