Incremental Change

Prefer Small Wins to a Big Bang

Big-bang and waterfall approaches try to tackle too much at one time. Even assessing the full inventory can become difficult. Starting small produces more accurate predictions, faster successes, lower risk of failure and overruns, and less burnout.

Large projects with vast objectives commonly overrun expectations or fail to meet their original goals. A better starting point is one job within a larger application, preferably a relatively low-level batch job. That gives the team a concrete production workload and the names of the programs actually being run.

A Practical Bottom-Up Process

1. Set up a git repository for the JCL file by cloning the sample proof-of-concept repository on COBOLworx GitLab.
2. Find the source files and place COBOL source in ./cob and non-COBOL source in ./other.
3. Compile each program with gcobol using the -find-copies and -report-call-with-variable-name options.
4. Find the required copybooks and place them in ./copy.
5. Create a Makefile using the template provided in the repository.
6. Run make and identify what fails and why.
7. Work with the COBOLworx Support Team to get everything compiled.
8. Test.

Expand Only After the Process Works

That outline, repeated JCL by JCL, is the minimalist approach. Keeping the process simple makes the learning curve easier. After one or two jobs have been completed successfully, the same approach can expand to larger parts of the application.

Incremental, bottom-up modernization can even begin as a side project to establish whether a broader effort is worth pursuing. Success on a small, well-defined workload creates evidence, confidence, and a much stronger basis for requesting additional resources.