A longitudinal retrospective covering twelve years of attempts to make SQLite distributed, examining projects ranging from early clustered forks to modern shared-nothing approaches. The author reflects on why the problem is harder than anticipated — SQLite's single-writer, file-locking architecture resists naive replication — and which design choices ultimately proved viable for read-heavy vs write-heavy workloads.
Background
SQLite is the most widely deployed embedded database, but its single-process, file-locking design was never intended for multi-writer distribution. Numerous projects over the years have attempted to extend it, with mixed results.
- Source
- Hacker News (RSS)
- Published
- Oct 24, 2026 at 09:16 AM
- Score
- 6.0 / 10