How are Map states priced in AWS Step Functions?

Question

I found the pricing documentation ambiguous with regard to the Map states that fan out based on an array in the input. Does anyone know if each fan out ends up being counted as a "state transition" incurring the $0.025 cost? Here is an example input and state machine definition for reference.

Input:

{ "data": [
        // Is each of these going to be a "state transition"?
        { "name": "a" },
        { "name": "b" },
    ] }

Definition:

{
  "StartAt": "Start",
  "States": {
    "Start": {
      "Type": "Map",
      "ItemsPath": "$.data",
      "End": true,
      "Iterator": {
        "StartAt": "Monitor",
        "States": {
          "Monitor": {
            "Type": "Task",
            "Resource": "some-lambda",
            "End": true
          }
        }
      }
    }
  }
}

Answer 1

The cost of Step Functions State Transitions is:

$0.000025 PER STATE TRANSITION THEREAFTER

$0.025 per 1,000 state transitions

With AWS Step Functions, you pay for the number state transitions. So let's see how many state transitions will Map State creates.

AWS Step Functions only charge customers for events that ends with Entered

For each Map State we have at least these 4 state transitions:

• MapStateEntered (Counted as state transition)
• MapStateStarted (Not Counted)
• MapStateSucceeded (Not Counted)
• MapStateExited (Not Counted)

And for each iterations of map state we have these 2 state transitions:

• MapIterationStarted (Not Counted)
• MapIterationSucceeded (Not Counted)

So for a Map State we can assume the cost is defined by:

cost = (1 + iterations * (steps inside iteration) ) * $0.000025

So for your example (An execution with a Map state with 2 iterations), the overhead of Map State is:

transitions: 1+2*1 = 3

cost: 3 * 0.000025 = $0.000075