pg_durable runs fault-tolerant long-running workflows as native SQL functions with built-in retry, checkpointing, and recovery—removing the need for separate job queues, schedulers, and orchestrators.
Summary
Developers can eliminate infrastructure glue (cron jobs, message queues, external orchestrators) by pushing workflow state directly into PostgreSQL. Execution survives crashes and resumes from checkpoints automatically, reducing distributed system complexity.
Why it matters
Developers can eliminate infrastructure glue (cron jobs, message queues, external orchestrators) by pushing workflow state directly into PostgreSQL. Execution survives crashes and resumes from checkpoints automatically, reducing distributed system complexity.
Implementation verdict
Replaces Temporal, external job schedulers, and async task queues for Postgres-native workloads. Requires PostgreSQL, Rust-based background worker, and DSL learning (~> and |=> operators). Ready now for vector pipelines and scheduled maintenance; production use depends on your tolerance for early-stage extensions.
Sources
Dev Signal
Get briefs like this in your inbox — free, every weekday.
100+ sources compressed into one 4-minute read. Ranked, cited, implementation-ready.