This implementation uses a Lambda map function (using a Distributed Map state from Step Functions) and a Lambda reducer function. The reducer function performs a final aggregation and writes the results to DynamoDB.

The reducer function is necessary because two child workflows in the Distributed Map run may process and find a high temperature for the same day. For example, child worflow 1 may find that Seattle, Washington, USA had the highest temperature on 2022-07 (July, 2022) while child workflow 2 finds that Jahra, Kuwait had the highest temperature on 2022-07. The reducer function will take a final pass through the outputs from all of the child workflows to find the correct highs.