╱ ╲
═══▶▕   ▏═══▶
     ╲ ╱

The actual process to be executed.

The workload can be a simple Lambda function, Docker via AWS Batch (or Batch array), or any other service like AWS Sagemaker (for training and validation), AWS Glue (for catalog maintenance) and AWS Athena (for chaining CTAS/INSERT INTO queries instead of chaining custom processing).

When a new lot of data arrives in a dataset, an arc is notified, which in turns starts a workload running against the newly arrived data.

A workload is declared by the appropriate module in the project file.

To run a custom Docker image, declare the arc as having the type aws:core:batchExecArc.

This will run the referenced Docker image within the AWS Batch service.