Deltas and Differences
A difference file comes from comparing two files. Trying to maintain all of the differences between two files can become very cumbersome. This kind of file can become very large and so the question is why go to such extremes? One should have a strong reason for creating a difference file. Such a file would need to have indicators specifying each of the relationships: per record:
Student Financial Services(SFS) maintains a dynamic list of all meal plans of students. The cafeteria has a separate program on a separate computer; one might ask why. See Buy vs Write. Daily the cafeteria needs the new list. On Saturday, the cafeteria loads the entire file; during the week, the cafeteria loads records from a difference file. In particular, the adds, the drops, and the changes. The entire file takes too long to load during the week. The SFS system does not generate a delta file, so a program compares the previous list to the current list and generates difference file.
The University contracts with an outside service to send notices to lenders. At the beginning of each semester, the University sends a complete file to the service. Every week, the University sends date records to the service provider. The service charges a small fee to send each notice. To save money, the University only sends adds, drops, and changes. Sending a same record would trigger a notice and the fee, hence, the University does not send a record when no changes have occurred.
A delta file comes from comparing two versions of the same file. In theory, the program that changes the file could also write records to a log or delta file. Keeping changes somehow requires before and after values. Tracking deletes requires some indicator and date. Likewise, inserts require a date. For many purposes, the delta file method is less complex than a difference file..
In some shops, a delta file is a measure of processing during time between the files.
- same record in both files
- found in first but not in second
- found in second but not in first
- same keys but different data in the files
Student Financial Services(SFS) maintains a dynamic list of all meal plans of students. The cafeteria has a separate program on a separate computer; one might ask why. See Buy vs Write. Daily the cafeteria needs the new list. On Saturday, the cafeteria loads the entire file; during the week, the cafeteria loads records from a difference file. In particular, the adds, the drops, and the changes. The entire file takes too long to load during the week. The SFS system does not generate a delta file, so a program compares the previous list to the current list and generates difference file.
The University contracts with an outside service to send notices to lenders. At the beginning of each semester, the University sends a complete file to the service. Every week, the University sends date records to the service provider. The service charges a small fee to send each notice. To save money, the University only sends adds, drops, and changes. Sending a same record would trigger a notice and the fee, hence, the University does not send a record when no changes have occurred.
A delta file comes from comparing two versions of the same file. In theory, the program that changes the file could also write records to a log or delta file. Keeping changes somehow requires before and after values. Tracking deletes requires some indicator and date. Likewise, inserts require a date. For many purposes, the delta file method is less complex than a difference file..
In some shops, a delta file is a measure of processing during time between the files.