A proposal to change a routine's algorithm has no profiling data behind it and no knowledge of the workload the routine runs against. Why is this risky?
Select an answer to reveal the explanation.
Short Explanation
Without profiling or workload data, you're guessing. You might speed up a rare path and miss the real hotspot, or make common cases slower.
Full Explanation
The correct answer is the option stating that a change made without workload context can optimise a rarely executed path while leaving the actual hotspot untouched, or improve one case while degrading a more common one. Without data on how the routine is exercised, a change is a guess, and guesses tend to optimise the easiest-to-picture case rather than the one that dominates real usage. The claim that any algorithm change automatically breaks compatibility with existing callers is incorrect because a well-designed change can preserve its interface. The claim that algorithms cannot be changed once a routine has been transformed from COBOL is incorrect because transformed routines can still have their algorithms revised. The claim that workload context is only needed for database access, not algorithms, is incorrect because workload context matters for any performance change. For the exam, require profiling or workload data before changing an algorithm.