Building a Workflow Executor on Postgres

I’ve posted the first half of this series before, which introduced workflow concepts and the front end builder.

This is the second part, which focuses on the infrastructure that drives the workflow executor.

Whenever you’re building an executor, you’ll face challenges like preventing concurrent execution, supporting incremental retries, and deciding on a storage model.

This shows how you can achieve that on top of Postgres, with minimal effort. I’ve used this pattern in many projects, and think it’s a solid answer to several problems if you know how to handle the edge cases.