Global multi-leader database
Three regions, each accepting local writes, meshed with async replication — version vectors detect conflicts and the LWW hazard is spelled out on the canvas.
What this design does
Multi-leader replication buys local write latency in every region and survives region loss — at the price of conflicts. This diagram shows the full-mesh async replication between three regional leaders, version vectors doing conflict detection, and the annotation every team needs taped to the wall: last-writer-wins silently drops data when clocks disagree. Import it and argue about your merge strategy with the actual topology in front of you.
The patterns it composes
- Multi-leader / multi-master mesh — every region accepts writes; replication is N-way async
- Cross-region async replication — local commits return fast; convergence lags by RTT
- Version vectors — concurrent updates are detected as conflicts, not races
- Last-writer-wins (hazard) — the default that quietly loses writes under clock skew
- Application-defined merge — the safe alternative: domain logic reconciles siblings
Import it, attach your own numbers, and verify the design against failures — or read the engineering notes.