GraphQL vs REST: When Each One Wins

GraphQL vs REST: The Honest Trade-offs

GraphQL isn't better than REST. REST isn't better than GraphQL. Each has trade-offs that make it better for specific situations.

Quick Comparison

Feature	REST	GraphQL
Endpoints	One per resource (`/users`, `/posts`)	Single endpoint (`/graphql`)
Data fetching	Server decides shape	Client decides shape
Overfetching	Common (get entire object)	Solved (request only needed fields)
Underfetching	Common (multiple requests)	Solved (nested queries)
Caching	HTTP caching (built-in)	Requires client-side caching (Apollo)
File uploads	Native (multipart)	Awkward (spec extension needed)
Learning curve	Low	Medium-High
Tooling	curl, Postman, any HTTP client	Apollo, Relay, GraphiQL

REST Example

// Get user profile + their posts + each post's comments
// Requires 3+ requests:
GET /users/42                    // user data
GET /users/42/posts              // user's posts
GET /posts/101/comments          // comments for post 101
GET /posts/102/comments          // comments for post 102
// N+1 problem: 1 + 1 + N requests

GraphQL Example

// Same data, one request:
query {
  user(id: 42) {
    name
    email
    posts {
      title
      publishedAt
      comments {
        text
        author { name }
      }
    }
  }
}

When to Choose REST

Simple CRUD apps
Public APIs (easier to document, cache, rate-limit)
File upload/download heavy services
Team is familiar with REST, deadline is tight
Microservices communicating with each other
When HTTP caching is critical

When to Choose GraphQL

Complex UIs with many related data types
Mobile apps (bandwidth matters, fetch only what you need)
Multiple client types (web, mobile, TV) with different data needs
Rapid frontend iteration (don't wait for backend to add endpoints)
When you'd otherwise build 20+ REST endpoints

GraphQL Gotchas

N+1 queries on the server: Use DataLoader to batch database queries
No HTTP caching: All requests are POST to /graphql — CDN caching doesn't work out of the box
Security: Malicious queries can be deeply nested. Use query depth limiting and complexity analysis
File uploads: Not part of the spec. Use multipart upload extensions or separate REST endpoints

The Middle Ground

Many teams use both: REST for simple CRUD and public APIs, GraphQL for complex frontend data requirements. They're not mutually exclusive.

Format your queries: JSON Formatter — format GraphQL responses for debugging.