A set of AWS Glue jobs (orchestrated by an AWS Glue workflow) produce data used by Amazon SageMaker Pipelines. The Glue jobs can run on a schedule or be started manually. You must integrate these Glue jobs into the SageMaker pipelines with the least operational overhead. Which approach satisfies this?

Callback steps in SageMaker Pipelines are designed for integrating external processes that run asynchronously and report their status back to the pipeline. This approach allows the SageMaker pipeline to trigger the AWS Glue workflow (which orchestrates the Glue jobs) and then pause, waiting for a callback token to resume. This effectively integrates the Glue jobs into the pipeline's flow with minimal operational overhead, as SageMaker manages the waiting and resumption. Using AWS Step Functions would introduce an additional orchestration layer, increasing complexity. Processing steps in SageMaker Pipelines are for running SageMaker-managed processing jobs, not directly invoking external Glue jobs. Amazon EventBridge can trigger actions but doesn't inherently provide the pausing and waiting mechanism needed for synchronous integration within a pipeline's execution flow.