Firestore — Serverless NoSQL document database

Firestore is a NoSQL database that stores JSON-like documents grouped into collections. Instead of rows and tables with fixed columns, you save flexible JSON objects and query them by field.

Its superpower is real-time: your mobile app subscribes to a query, and every time a document changes — anywhere in the world — all subscribed clients receive the update instantly. No WebSocket to code, no polling. It's how chat apps, collaborative editors, and live scoreboards are built with minimal backend.

Offline support comes built in (in the Firebase SDKs): the app works without internet, queues writes, syncs everything when back online — conflicts handled automatically.

Security rules let you authorize directly from the database layer (no backend in the middle): 'a user can read their own docs', 'only admins can delete'. Written in a dedicated language, evaluated on every request.

When NOT to use Firestore: complex SQL joins (no joins, data must be denormalized), heavy analytics (go to BigQuery), strong relational transactions across many docs (limited compared to Postgres). For typical app workloads — user profiles, feeds, chats, settings — it's excellent.

Comparison: Firestore = MongoDB Atlas = AWS DynamoDB (DynamoDB is closer in model) + Firebase Realtime DB. Firestore is the evolution of both Datastore and Firebase Realtime DB into one product.