Why TigerBeetle is the most interesting database in the world
20 hours ago
- #zig
- #database
- #distributed-systems
- TigerBeetle is a unique database that does the opposite of conventional databases, focusing on writing code slowly, deterministic simulation testing, and having zero dependencies.
- It enforces static memory allocation, keeps assertions enabled in production, and uses Viewstamped Replication and Zig instead of Raft and Rust.
- TigerBeetle is designed for financial transactions, treating debits and credits as first-class primitives, offering high performance with 8,190 transactions packed into a single 1MiB query.
- The database is distributed by default, handling storage faults and clock synchronization issues, and uses Protocol Aware Recovery for fault tolerance.
- Built entirely in Zig, TigerBeetle benefits from static memory allocation and a great developer experience.
- It employs Deterministic Simulation Testing (DST) via the VOPR cluster, simulating years of runtime in days to catch rare conditions.
- TigerBeetle's engineering methodology, TigerStyle, emphasizes simplicity, elegance, and the use of assertions, inspired by NASA's Power of Ten.
- The database is Jepsen-proof after just 3.5 years, showcasing its robustness and innovative approach to database design.