Fireship
June 19, 2026
Turso: Rewriting SQLite in Rust with Concurrency and Vector Search
TL;DR
Turso is a Rust-based rewrite of SQLite that adds modern features like concurrent writes, async operations, and native vector search while maintaining backward compatibility with the world's most trusted database.
Key Takeaways
- 1SQLite is the most widely deployed database in the world, powering billions of devices from iPhones to Mars rovers, but its single-threaded architecture and closed contribution model motivated a complete rewrite
- 2Turso adds three major features SQLite lacks: concurrent multi-writer support for simultaneous database access on different rows, async operations that don't block threads, and native vector search for AI embeddings
- 3The project maintains backward compatibility as a drop-in SQLite replacement while using deterministic simulation testing to prevent data loss by injecting failure scenarios and replaying them from random seeds
- 4Turso is truly open-source accepting community contributions, addressing a key limitation of SQLite's closed three-person maintenance model
Notable Quotes
“The most trusted and resilient software out there”
— Narrator, describing SQLite
“You're only viewed as an idiot until it works”
— Narrator
“Multiple writers can work on different parts of the database simultaneously and only conflict if they actually touch the same rows”
— Narrator, explaining Turso's concurrency feature
Chapters
1. The Origin and Dominance of SQLite
D. Richard Hipp created SQLite in 2000 to solve a Navy destroyer application problem by embedding a database directly into applications. Over 25 years, it became the most trusted software ever written, with trillions of instances across all devices.
2. Why Rewrite SQLite in Rust?
Two accomplished engineers (one wrote a latency book, the other a top Linux kernel contributor) decided to rewrite SQLite in Rust to address its limitations: it lacks true open-source contribution model and doesn't support modern features needed by contemporary applications.
3. Turso's Key Features
Turso introduces concurrent multi-writer access to different database rows, async operations that return control without blocking threads, and native vector search with indexing for AI embeddings—solving problems that typically require bolting on separate databases.
4. Building Trust Through Testing and Compatibility
Turso achieves reliability through deterministic simulation testing that runs the database in a simulated universe and injects failures like power loss and disk corruption. It maintains full backward compatibility as a SQLite drop-in replacement.
5. The Open-Source Advantage
Unlike SQLite's closed three-person maintenance model, Turso embraces traditional open-source contributions, allowing the community to improve the database and potentially earn the 25 years of trust SQLite has built.
Key People & Entities
- D. Richard Hipp
- Creator of SQLite, spent 25 years building the world's most trusted database
- Turso founders
- Two engineers rewriting SQLite in Rust; one authored a latency book, the other ranked as top-five Linux kernel contributor by Linus Torvalds
- JetBrains
- Sponsor providing Juny, an AI coding agent that optimizes token costs by selecting appropriate AI models for different tasks
Glossary
- Drop-in replacement
- Software that can be substituted for another without requiring changes to the existing code that uses it
- Deterministic simulation
- Testing method that runs software in a controlled simulated environment and replays scenarios using the same random seed to reliably reproduce and fix bugs
- Vector search
- Database functionality to store and query embeddings (numerical representations of data) to find the closest or most similar items
- Embeddings
- Numerical representations of text, images, or other data used by AI models to understand semantic meaning and relationships
- Async operations
- Non-blocking operations that return control to the calling program while the task completes in the background