I have recently been thrown into a situation where I'm dealing with financial data (batch processing environment) that is transferred on a daily basis from a Mainframe COBOL system to (after some intermediate steps including ETL) an Oracle system. As I am not in the IT department, all I really have direct access to is the dollars the Mainframe systems people say they transferred and the dollars the Oracle systems people say they received. Those dollars may include multiple batches. As ridiculous as this may seem, whenever there is a problem (for example, there is a multi-day delay in receiving a particular day's work in Oracle), the way the Oracle people are going about trying to find the data (that came into Oracle late) is by searching Oracle batch reports for a batch or group of batches whose dollars add up to (or closely add up to) the amount the Mainframe people say they transferred. The batches (batch numbers) I am referring to are on the Oracle side only. They are not batch numbers that are transferred by the Mainframe and captured by Oracle. So they are of little if any use in cross-referencing between the two systems.
My question is: What identifying data should be being transferred by the mainframe to Oracle to enable the Oracle IT department to identify which mainframe data was attempted to be transferred, what data actually got into Oracle and what data didn't get in? The entire problem is currently the inability to match data that was sent with data that was received (and also what data was attempted to be sent but failed). Thank you.