Identifying the processing unit is the key to the design of streamed batch programs. The aim here is to identify the smallest possible unit of work that can be executed without risk of overlap between processing units. While there may be a readily identifiable grouping (by case or participant), more work will be needed where some serial processing may be required. It's important to note that this sub-division of the processing is a fundamental activity where the work of a batch program needs to be done in parallel and this is not merely a feature of the batch streaming infrastructure, but of the problem being addressed.