Step Functions

When a piece of work spans several steps — call a service, wait, branch on the result, retry on failure — something has to hold the state between them. AWS Step Functions is the managed way to do that. The workflow is defined as a state machine, and the service tracks progress, retries and error handling, so the application code does not have to.

Step Functions covers the two workflow types, Standard and Express. They differ in execution guarantees, duration and cost, so the choice between them matters early. The page also covers the state types and limits.

Not every coordination problem needs a state machine. Step Functions and Batch compares Step Functions with SQS and AWS Batch, and explains how Batch compute environments are put together for large volumes of independent jobs.