SurrealDB — Scalable Multi-Model Database for Modern Apps

What it is

SurrealDB is a scalable, multi-model database designed for the real-time web. It combines relational, document, graph, and time-series data models in a single database with its own query language called SurrealQL. You can write SQL-like queries that traverse graphs, filter documents, and aggregate time-series data in one statement.

SurrealDB targets full-stack developers building real-time applications who want to avoid managing separate databases for different data models.

How it saves time or tokens

SurrealDB eliminates the need for a document store, a graph database, and a relational database. One connection string, one query language, one backup process. The built-in permissions system means you can expose SurrealDB directly to the frontend with row-level security, removing the need for a separate backend API for simple CRUD operations.

Real-time subscriptions via WebSocket push data changes to clients without polling.

How to use

1. Install SurrealDB: curl -sSf https://install.surrealdb.com | sh
2. Start the server: surreal start --user root --pass root
3. Connect with the CLI: surreal sql --conn http://localhost:8000 --user root --pass root
4. Create tables and query with SurrealQL

Example

-- Create a user document
CREATE user:alice SET name = 'Alice', age = 30, role = 'engineer';

-- Create a graph edge
RELATE user:alice->works_at->company:acme SET since = '2024-01-01';

-- Query with graph traversal
SELECT name, ->works_at->company.name AS company
FROM user
WHERE role = 'engineer';

-- Real-time subscription
LIVE SELECT * FROM user WHERE age > 25;

Related on TokRepo

• Database tools -- Database management and query tools
• Coding tools -- Full-stack development tools

Common pitfalls

• SurrealDB is younger than PostgreSQL or MongoDB; the ecosystem of drivers and ORM integrations is smaller
• The direct-to-frontend pattern requires careful permission configuration; misconfigured DEFINE TABLE PERMISSIONS can expose data
• SurrealQL looks like SQL but has significant differences in syntax for graph traversals and record IDs