Overview

hook-sync

SQLite replication that just works. Multi-server, multi-writer, multi-runtime. Zero data loss.

SQLite is the fastest SQL database in the world — zero config, single file, serverless. But it can’t replicate. Until now.

hook-sync adds replication to SQLite via triggers + HTTP sync. No consensus algorithm. No Raft. No coordinator. Just triggers, ACK, and UUID. The result: 3.8x faster than Postgres at batch 10K, with multi-writer active-active, crash recovery, and split-brain safety.

Why hook-sync?

Problem	Postgres	Litestream	rqlite	hook-sync
Replication	✅ WAL streaming	✅ WAL to S3	✅ Raft consensus	✅ Trigger + ACK
Multi-writer	❌ Primary-only	❌ Read-only replica	❌ Leader-only	✅ All nodes write
Zero write penalty	❌ WAL sender overhead	✅ Async	❌ Raft quorum per write	✅ Async, 0% overhead
Cross-runtime	❌ C server only	❌ Go only	❌ Go only	✅ Go, Bun, Node interop
Split-brain safety	✅ Sync replication	❌	✅ Raft	✅ Timestamp LWW
Crash recovery	✅ WAL replay	✅ S3 restore	✅ Raft log	✅ _changes replay
Speed (batch 10K)	8,278 QPS	N/A	N/A (Raft quorum)	31,558 QPS

hook-sync occupies a gap no other project fills: SQLite simplicity + multi-writer replication + cross-runtime interop, without consensus overhead.

Quick Start

Get up and running in 5 minutes with Go, Bun, or Node.

Libraries

Importable packages for Go and JS — trigger-based or hook-based capture.

Topologies

Point-to-point, full mesh, dedicated hub, multi-region.

Wire Protocol

Language-agnostic protocol spec — implement in any language.