Extra processing that is required is processed by the chunker after the chunking has been completed. This allows processing to start on other streams. However the extra processing happens before the chunker starts it's own stream (if configured to do so), in a single database transaction. It is important therefore to take the total volume of processing expected when designing any extra processing into account. In general if the five year maximum volume of data is likely to take more than a couple of minutes to process, then a separate batch program (see Multiple Batch Programs) is a better option than using extra processing. Where extra processing is typically used is where there is some additional processing cannot be associated with any one chunk.