Yesterday I came upon an article about some guys saving an IBM 360 Model 20 mainframe from the 1960’s. It seems appropriate to think about the system architecture which would have worked on that machine at the time, and it’s origins. The system would be been based upon tabulating systems.
If today’s financial systems, and the tabulating System principles upon which they are based, were simple, no one would call them spaghetti code. But because the principles have been forgotten by many, they appear more complex.
How to come to understand them?
One approach that has been successfully used numerous times is called “show me the money.” It focuses on the data used, as inputs, and outputs, to discover the tabulating System patterns in the systems.
From two weeks ago, remember that the posting process is a major pillar in financial and tabulating systems. It has three primary files:
- Input transactions
- Input master file updated in last process execution
- Output master file
Data discovery is the process of looking at the data in the files and trying to detect:
- Sort Order
- Common keys
- Monetary value fields
- Accumulated balances
If one can find a sample case of a few transactions which accumulate to a single balance, which has been updated between the input and the out balances, then code has been cracked. So many other functions will hang off this basic process.
Next Up: Metadata in Tabulating Systems
Previous: Complexity in Mainline Routine