The purpose of this guide is to describe how to design and develop streamed batch programs using the infrastructure provided by the application.
A streamed batch program is one where the processing load is divided into streams of independent processing, where these streams can be processed on separate machines as required. At a technical level the work of the batch program is divided into processing units which can be processed independently of each other, for example the payments for a person or the reassessment of a case.
At execution time, there is "chunker" process which identifies these processing units and clusters these into chunks of a predefined size. The stream processes then process one chunk at a time, once all the chunks have been processed the chunker summarizes the processing carried out by the streams and reports on the batch processing carried out.