One of the problems associated with large data base systems is the lingering death associated with corrupted data bases. Unfortunately, the program that first discovers the corruption is usually the victim and not the perpetrator. Two tedious processes then ensue. First, the result must be corrected, and, second, the cause must be discovered. This latter problem is very complex in that usually the perpetrator is long since gone and has left few to no clues behind. Some suggestions:

1. Put in early alarm systems to detect some problems as they occur and try to contain them; for example, validity check all records as they are entered into the data base.
2. Instrument the system to keep clues, for example, store the program ID and date of update.
3. Reduce the potential problems by consolidating data base access routines which handle inverted files.