LumoSQL
lastmod 2025-06-03created 2025-06-031 edits
Project for extending SQLite features and providing alternative storage backends
Features
- A project that adds security, privacy, performance, and instrumentation features to SQLite; it uses dynamic source combination rather than forking, making it easy to track changes against upstream
- Does not replace SQLite directly; only the needed features can be applied to SQLite individually (good for gradual adoption/extension)
- Makes up for the limitations of existing SQLite and meets privacy requirements such as GDPR
- Runs combined benchmarks across various SQLite/LMDB/BDB versions and data size (DATASIZE) options, automatically storing results in a DB
- Each benchmark run is permanently identified by a SHA3 hash, making data analysis and reproducibility easy
- Automates builds, combinations, and benchmarks based on the not-forking Perl tool, storing results in a DB
- Previous alternative SQLite backends were mostly hard forks, but LumoSQL's key differentiators are being non-forked, dynamically combined, and multi-backend
- Started from Howard Chu's 2013 experiment porting to LMDB; benchmarks have since shown that SQLite's own performance has also steadily improved
- Going forward, expanding its role as a platform for experimenting with various KVS architectures, attempting upstream contributions, and experimenting with advanced security/integrity/privacy features