Question 700
The service owner asks how to avoid a repeat of: A full database snapshot reports equal row counts but several customer balances differ after migration. Which operational check is the best match?
Explanation
Correct Answer
Include semantic data-value checks alongside row-count validation. The original failure arose from: Row totals alone cannot establish data-value equivalence. The control reduces that failure mode.
Incorrect Answers
This would record source and target checkpoint positions with validation reports. It is useful when a parallel-run comparison flags mismatches because the same transaction is captured at different times in source and target snapshots, but it does not prevent the issue here: row totals alone cannot establish data-value equivalence.
For the situation in which a migration runbook declares cutover safe whenever lag reaches zero but does not verify application writes are paused, it makes sense to require an auditable final write and log position before cutover. Here, row totals alone cannot establish data-value equivalence, so the safeguard misses the recurring fault.
This would assess engine-specific code and dependencies at migration discovery. It is useful when an Oracle modernization project targets a PostgreSQL engine without an inventory of PL/SQL packages and stored procedures, but it does not prevent the issue here: row totals alone cannot establish data-value equivalence.