Petri – Per-Test Database Forking for Parallel Test Execution

Developers using Postgres struggle with slow test suites due to database locking and mocking complexity. Petri is a drop-in Postgres image with a proxy that automatically forks a database for each test connection, enabling true parallelization. Target users are backend engineers and QA teams.

The pain is real and well-documented — teams running Rails, Django, or any ORM-heavy stack on Postgres routinely hit serialization bottlenecks in CI, and the shift toward faster feedback loops (driven by GitHub Actions pricing and trunk-based development) has made slow test suites a budget line item. No clear incumbent owns this specific niche; PlanetScale's branching is the closest conceptual relative but targets production workflows and is MySQL-only, leaving the Postgres parallel-test space largely to home-rolled fixtures and jest-style isolation hacks. The $1k–5k/mo revenue band is realistic but tight — this is a per-seat or per-seat-equivalent devtools play where individual engineers rarely expense tools without team buy-in, so the sales motion is bottoms-up but conversion depends on champions getting budget approval. The biggest risk is that Postgres itself closes the gap: logical replication improvements and copy-on-write extensions like `pg_copy` are active areas of upstream development, and a sufficiently good native solution would undercut the entire value proposition before any meaningful customer base is locked in.