PGSimCity: an explorable 3D city that visualizes PostgreSQL internals
PGSimCity visualizes PostgreSQL internals as an explorable 3D city, covering WAL, checkpoints, autovacuum, buffers and replication in real time.
PGSimCity is an independent, non-commercial educational project that turns a running PostgreSQL cluster into a walkable, breakable 3D city. Districts map to real internals — backend processes, the shared_buffers pool, WAL, checkpointing, autovacuum and replication — with every building's behavior tied to actual Postgres formulas such as the WAL trigger point, cache hit ratio and the clock-sweep usage-count cap.
It targets engineers who are strong technically but have never operated a database, letting them see why checkpoints spike latency, how a forgotten open transaction bloats tables indefinitely, and what synchronous_commit actually costs. Built-in scenarios like cache thrash, long-running transactions, checkpoint storms and slow replica replay make these effects observable in real time rather than theoretical.
Rather than running actual PostgreSQL source, the project ships a scaled model validated against official documentation through multiple expert review rounds and backed by a 234-test suite. It runs entirely in the browser, built with three.js, TypeScript and Vite, with no install and no framework dependency.
This synthesis was produced from its source by AI; there is no human editor or manual review step. How we work