APIs — REST vs GraphQL
APIs are the backbone of communication between clients and servers.
In system design, you’ll often be asked: “Would you use REST or GraphQL?”.
Both are widely used, but they solve different problems.
1. REST (Representational State Transfer)
- Resource-based API style.
- Endpoints map to resources (
/users,/orders,/products). - Uses HTTP verbs:
GET,POST,PUT,DELETE. - Data usually in JSON.
Pros
- Simple and widely adopted.
- Easy to cache (HTTP caching, CDNs).
- Strong separation of concerns.
- Mature tooling (Postman, Swagger, OpenAPI).
Cons
- Over-fetching: client may get more data than needed.
- Under-fetching: may require multiple API calls to get all data.
- Fixed endpoints, less flexible.
Best for
- CRUD-style applications.
- Systems where caching is important (e.g., e-commerce product APIs).
- Simpler architectures with well-defined resources.
2. GraphQL
- Query language for APIs, developed by Facebook.
- Single endpoint (
/graphql). - Clients specify exactly what data they need.
- Strongly typed schema.
Pros
- No over-fetching or under-fetching — client gets exactly what it requests.
- Great for mobile/slow networks (efficient data transfer).
- Strong typing improves client/server contract.
- Good for aggregating data from multiple sources.
Cons
- Harder to cache (queries are dynamic).
- More complex to implement.
- Can result in expensive queries if not rate-limited.
- Tooling still evolving compared to REST.
Best for
- Complex applications with diverse clients (web, iOS, Android).
- Systems where bandwidth efficiency is important.
- Aggregating data from multiple backends.
3. REST vs GraphQL Comparison
| Feature | REST | GraphQL |
|---|---|---|
| Data fetching | Fixed endpoints, may over/under-fetch | Flexible, exact data |
| Number of calls | Often multiple | Usually one |
| Caching | Easy with HTTP/CDN | Harder, needs custom logic |
| Schema | Implicit (docs-driven) | Explicit, strongly typed |
| Complexity | Simpler to implement | More complex to implement |
| Tooling | Mature (OpenAPI, Swagger) | Improving (Apollo, Relay) |
4. Hybrid Approaches
- Some companies use both.
- REST for simple resources (good caching).
- GraphQL for complex data aggregation.
Example:
- Twitter uses REST for most APIs, GraphQL for some modern clients.
5. Interview Tips
- Don’t say “GraphQL is always better.” Show trade-offs.
- If caching/CDN is critical → REST.
- If flexibility and mobile optimization matter → GraphQL.
- Mention security concerns (GraphQL queries can be too deep → need query depth limiting).
- Say: “I’d start with REST for simplicity, but if clients need flexible queries, I’d consider GraphQL.”
6. Next Steps
- Learn about Load Balancing.
- Explore CDNs & Edge Computing.