The GraphQL N+1 Problem & DataLoader Solution

In GraphQL servers, each field in a query is resolved by an independent, isolated resolver function. The GraphQL execution engine traverses the Abstract Syntax Tree (AST) top-down and executes field resolvers concurrently.

The Problem Walkthrough:

Consider this simple query fetching 100 posts and their authors:

graphql
Copy
query GetFeed {
  posts(limit: 100) {
    id
    title
    author {
      name
      avatarUrl
    }
  }
}

The Execution Cascade:

1. The top-level Query.posts resolver runs 1 SQL query:

   sql
   Copy
   SELECT * FROM posts LIMIT 100;

2. For each of the 100 posts returned, the nested Post.author resolver fires independently:
   • Post 1: SELECT * FROM users WHERE id = 12;
   • Post 2: SELECT * FROM users WHERE id = 45;
   • ...
   • Post 100: SELECT * FROM users WHERE id = 88;

Total Database Queries: 1 + 100 = 101 queries for a single HTTP request!

If each post also requested 5 comments and comment authors, the total query count explodes to:

1 + 100 + (100 × 5) + (500 × 1) = 1,101 database queries!

This overwhelms database connection pools, causes thread starvation, and spikes API latency to multiple seconds.