ITPEC FE Subject A April 2026 Question 23

Before and after values of each update, used for recovery — the log (journal) records every change so the DBMS can undo or redo it.

Every update writes a log record holding the before image and the after image of the data. After a failure, recovery uses them in two directions:

• •Rollback (undo) — writes back the before images of transactions that had not committed
• •Roll-forward (redo) — reapplies the after images of committed transactions, for example onto a restored backup after a disk failure

Answer (d)

Why not others:

• •(a) — keeping a live copy on another disk or at another site is mirroring or replication
• •(b) — copying the whole database disk by disk is a backup (dump)
• •(c) — writing updated buffers from main memory to disk to shorten recovery is a checkpoint

Key rule: A backup restores an old snapshot and a checkpoint shortens recovery, but only the log holds the individual changes, as before and after images, needed to undo or redo transactions.

Databases · AI-written, may contain errors