The best workflow engines, according to hackers
12 workflow engines recommended in Hacker News comments over the past year, ranked by how many separate people recommend each one.
- Most recommendedTemporal138 people recommend it
- Built to lastHatchetowners report a median of 2.4 years
- Least regretStepKit10 recommend, 0 criticize
- Free and lovedApache Airflow18 people recommend it
All workflow engines
- TemporalRecommended for reliable long-running workflows with built-in retries, persisted state and crash recovery; critics call it overkill for small tasks, heavy to self-host and pricey on Cloud.138recommend
- DBOSRecommended for durable workflows and queues on Postgres, with far less complexity than Temporal. Common caveat: scaling and executor elasticity sit behind a paid tier.39recommend
- Apache AirflowWidely used open-source engine for scheduled ETL pipelines, valued for Python workflows and its UI; critics say it fits recurring jobs better than durable or database-aware workflows.18recommend
- RestatePeople recommend it for durable workflows and agents, with recovery, retries and state built in. Main caveat: critics question its availability and single point of failure.16recommend
- WindmillPeople recommend Windmill as a dependable workflow engine that integrates well with Python, Postgres and DuckDB. The most common caveat is its licensing.11recommend
- Absurd WorkflowsRecommended as a minimal, simple durable workflow engine that leaves you full control over how it is used.11recommend
- StepKitOpen, cross-platform durable execution standard with an explicit API and pluggable backends; commenters like it for avoiding reinventing the wheel.10recommend
- InngestPeople recommend Inngest for durable retries and observability without touching business logic. The common caveat is that you give up transactional job creation.8recommend
- Argo WorkflowsKubernetes workflow engine people pick for long-running pipelines and self-hosted setups, especially if they already run Kubernetes. Critics say setup is complex and startup is slow.6recommend
- Cloudflare WorkflowsPeople recommend it for running long multi-step jobs that can last days without timeouts. Critics raise separate concerns about latency, deletion, and distributed use.6recommend
- HatchetOpen-source platform for background jobs, praised for simplifying stacks; critics say it hides concepts developers need.6recommend
- DagsterPeople recommend Dagster as an easy-to-use orchestrator for data workflows, citing its developer experience and suggesting it for orchestration problems.6recommend