How Figma’s databases team lived to tell the scale

Figma went from a single Postgres database in 2020 to a distributed architecture with vertical partitioning, read replicas, and a dozen vertically partitioned databases by the end of 2022. Scaling from vertical to horizontal sharding was a complex but necessary step to ensure continued growth. Instead of migrating to alternative databases or NoSQL solutions, Figma chose to build a customized horizontally sharded solution on top of their existing infrastructure. The key highlights included colocations, logical sharding, a custom DBProxy query engine, and full logical replication to avoid complex filtered replication. The unique approach taken by Figma allowed for scalability while minimizing downtime and application layer changes.

https://www.figma.com/blog/how-figmas-databases-team-lived-to-tell-the-scale/

To top