What is optimization work best organized as, once AI-assisted analysis has identified candidate hotspots in a mainframe application?
Select an answer to reveal the explanation.
Short Explanation
You don't want one giant rewrite you hope works. You want small, reviewable changes backed by regression tests.
Full Explanation
Optimization work is best organized as small, reviewable changes, each backed by regression tests, rather than one large rewrite. Optimization changes behaviour-preserving code in ways that can still introduce regressions, so each change should be small enough to review meaningfully and paired with tests that confirm it did not alter output. The combined rewrite that addresses every hotspot at once, reviewed after the fact, is wrong because it makes it impossible to attribute a regression to a specific change, and review after the fact catches problems too late. Changes applied directly with no separate review step, on the basis that the AI already validated them, are wrong because AI-suggested changes still need human review; the tool identifies candidates, it does not guarantee correctness. Grouping changes to minimize the number of deployments, regardless of how many hotspots each change touches, is wrong because it trades safety for convenience, which is the wrong trade in production code. For the exam, know that optimization is applied incrementally with review and tests at each step.