BriskDB: Transforming SQLite Shards into Distributed Databases with Rust Engine
August 15, 2026
BriskDB is a protocol-neutral Rust engine that turns multiple ordinary SQLite shard files into a distributed database with parallel writes, PostgreSQL compatibility, HTTP access, and embedded Rust and Python APIs.
Alpha-stage with current limitations: no general cross-shard atomic transactions, development loopback-only HTTP, partial global ordering, and a stop-gap backup approach of snapshotting the data directory on exits.
Key capabilities in place include durable routing across independent write-ahead logs, exact-key routing, bounded reads, HTTP query and write API, an admin data browser, loopback-only PostgreSQL wire protocol, TLS/SCRAM for remote access, and live metrics and health reporting.
Roadmap envisions adding more protocols and backends (MongoDB, MySQL, etc.) and features serverless storage, broader protocol support, while clearly labeling BriskDB as an alpha with honest boundaries.
The architecture dispatches operations via a router handling 4,096 virtual buckets to per-shard WALs, enabling multi-client access through web/HTTP, PostgreSQL, Rust embedding, and Python embedding.
Each shard remains an independent SQLite WAL file, allowing parallel writes without central write locks and preserving all existing SQLite tooling.
A practical demo and quick-start guide show installing via Python wheels, running a local multi-shard server, verifying health and PostgreSQL readiness, with the local data browser accessible at /admin using loopback credentials.
Shard-safe IDs are provided through two opt-in designs: native_range_v1 with non-overlapping 64-bit ranges and hilo_v1 that leases blocks of 4,096 IDs, with versions tracked in a manifest.
Documentation stresses using ordinary SQLite files (no forking), cross-shard pruning with global uniqueness, and a single engine governing protocol behavior across PostgreSQL, HTTP, Rust, and Python interfaces.
Summary based on 1 source
