Google has announced Cloud Firestore, a document database for mobile, web and server applications.
Last year we reported on Google Firebase, a database for mobile and web development, providing real-time and offline access to data and integration with many Google services. They follow now with a "brand new" offering called Cloud Firestore, a NoSQL database in the cloud for mobile, web and server applications. Its main features are: a flexible data model, complex querying, real-time data updates, support for offline access, scalability across multiple regions, strong consistency, batch operations, support for transactions, and integration with the rest of Google ecosystem, including Cloud Functions. Besides REST and RPC APIs, Firestore comes with SDKs for Android, iOS (Swift and Objective-C), JavaScript (including Node.js), Java, Python and Go.
It seems that with Firestore Google wants to address some shortcomings with Firebase, also called now Realtime Database. The main differences between the two solutions are:
- Data model. While Realtime keeps data in one JSON file, Firestore introduces the concept of documents organized in collections.
- Offline. Firestore adds offline support for web applications while Realtime supports only mobile ones.
- Querying. Due to a better data model, Firestore supports more complex and faster queries.
- Operations. Firestore can combine multiple operations in atomic batches, and transactions are repeated until completed.
- Zones. Realtime is limited to one zone in a single region while Firestore can span multiple regions.
- Scalability. Realtime is limited to 100K connections and 1K writes/sec. More can be achieved by sharding the database. Firestore automatically scales beyond those limits when needed.
- Security. Firestore has a simpler data access control.
Regarding the data model, Firestore keeps data in documents organized in collections. Each document contains key-value pairs, and can have associated subcollections. Beside the standard types -boolean, integer, float, byte, string, geo point, date and time, null – Firestore supports arrays and maps, the latest being a form of nested objects. The database is schemaless.
AngulaFire has announced support for Firestore with observable streams that can be integrated in Angular applications, including in Animations, Forms, and Routers. Firestore is currently in beta and there are some limits in using it. Also, pricing is different than the one used for Realtime.