eolas/neuron/22c4bb42-86c2-4906-9753-27655e15c1d9/Step_functions_and_state_machines.md

---
tags: [AWS]
---

# AWS Step functions and state machines

_AWS Step Functions_ is an AWS service that allows you to coordinate distributed
applications and microservices by creating workflows.

Workflows are made up of a series of steps and are known as **state machines**.
A state machine defines:

- the states of the workflow
- the transitions between states
- the inputs and outputs of each state

State machines are defined via a JSON object which specifies the states and the
transitions between them. It also includes error catchers and retry logic.

At the beginning you define a `StartAt` state which is the entrypoint of the
state machine. This can be manually triggered, or more likely, triggered by
another AWS service such as a
[Lambda](Lambda_programming_model.md), an
[API Gateway](AWS_API_Gateway.md) request or a messaging/queue
event.

The state machine ultimately ends at an end state. In between are various
intermediary states which can include:

- Task states (do something)
- Choice states (do something based on conditions)
- Wait states
- Success states (if preceding state succeeds do X)
- Fail states (if preceding state fails do Y)
- Parallel states

The diagram below shows the definition of a given state machine. On the left is
the JSON specification. On the right is a diagramatic representation that shows
the control flow and all possible steps that comprise a given pathway / state:

![](static/state-machine-definition.png)

For each execution of the state machine (each time it is triggered) you can
review the runtime. The flow diagram will highlight green to show the given
pathway, and you can also see the inputs and outputs for each step and any
errors:

![](static/state-machine-execution.png)