Categories
Self-Hosted Developer Tools Database SQL SQLite Vector Search Embedded Database ReplicationStars
Forks
Watchers
Developer links
libSQL
libSQL bridges the gap between SQLite's legendary simplicity and the server-accessible database model developers expect from PostgreSQL and MySQL, adding network access, replication, and vector search while maintaining full file format and API compatibility. The sqld (SQL daemon) component transforms SQLite into a network-accessible database server accepting queries via HTTP with JSON payloads, WebSocket connections, and a Hrana protocol optimized for low-latency edge access. Client libraries for TypeScript, JavaScript, Rust, Go, and Python connect to sqld identically to any traditional client-server database, while the LD_PRELOAD compatibility layer enables switching existing SQLite applications from local to remote mode without code changes. Embedded replicas synchronize a local SQLite copy inside your application process with the primary server, delivering sub-millisecond read latency while maintaining consistency through streaming replication. Bottomless replication continuously backs up database state to S3-compatible object storage with generation-based snapshots and WAL page shipping, providing point-in-time recovery without manual backup scheduling. Native vector search enables semantic similarity queries directly within SQL using cosine distance functions, eliminating the need for external vector database infrastructure. Read replica support distributes query load across multiple sqld instances, and integration with mvSQLite adds multi-version concurrency for high-availability deployments. Created by Turso with 17,100+ stars and active production deployment. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. MIT licensed.
Benefits
- SQLite Goes Client-Server
- The sqld daemon exposes SQLite over HTTP and WebSocket, enabling remote database access from any language with client libraries for TypeScript, Rust, Go, Python, and more.
- Embedded Replicas in Your App
- Synchronize a local SQLite copy inside your application process with streaming replication from the primary server, delivering sub-millisecond reads with automatic consistency.
- Bottomless S3 Backup
- Continuous replication to S3-compatible storage with generation-based snapshots and WAL page shipping provides automatic point-in-time recovery without manual backup scheduling.
- Full SQLite Compatibility
- Maintains identical file format and complete API compatibility with SQLite, allowing existing applications to adopt libSQL with zero schema migration or query modification required.
Features
- HTTP Query Interface
- Accept SQL queries as JSON payloads over HTTP, enabling database access from serverless functions, edge workers, and any HTTP-capable client without native drivers.
- Native Vector Search
- Execute semantic similarity queries using cosine distance functions directly within SQL statements, eliminating dependency on external vector database infrastructure.
- Read Replica Distribution
- Deploy multiple sqld instances as read replicas receiving streaming updates from the primary, distributing query load for high-throughput read-heavy workloads.
- LD_PRELOAD Compatibility
- Switch existing SQLite applications from local to remote database access without code changes using dynamic library preloading, providing zero-effort migration path.
- Hrana Protocol
- Low-latency WebSocket protocol optimized for edge access with connection multiplexing, pipelining, and batched transactions for efficient network utilization.