Queries vs mutations vs subscriptions in GraphQL?

Question

IT Interview Practice · Accepted Answer

They are the three **operation types**, each mapped to a root type in the schema:

- **Query** — reads data. Fields run **in parallel** and must have no side effects (like HTTP GET).
- **Mutation** — writes data (create/update/delete). Top-level fields run **sequentially**, so ordering is guaranteed.
- **Subscription** — a long-lived stream: the server pushes new data to the client over a transport like WebSockets whenever an event fires.

```graphql
query GetPost {              # read
  post(id: "1") { title }
}

mutation AddComment {        # write; returns the created object so the client updates its cache
  addComment(postId: "1", text: "Nice!") {
    id
    text
  }
}

subscription OnComment {     # live: fires every time a comment is added
  commentAdded(postId: "1") {
    id
    text
  }
}
```

Mutations take **arguments** (often an `input` object) and should **return the affected data** so the client can refresh without a second request. The sequential rule matters: two mutations in one document — `deposit` then `withdraw` — always run in that order, whereas two query fields might resolve concurrently.

## Why it matters

Interviewers check that you know **queries are parallel and side-effect-free while mutations are sequential**, and that a subscription needs a stateful transport (WebSocket/SSE). Returning updated data from a mutation is the practical detail that separates people who have shipped GraphQL from those who have only read about it.

Why it matters

Related questions