ITPEC FE Morning April 2022 Question 28

Lock Granularity in RDBMS — trade-offs between row-level and table-level locks.

Two transactions update multiple rows in the same table in parallel:

• •A = table — lock contention is more likely with a table-level lock because the entire table is locked, blocking even non-overlapping row updates
• •B = row — more RDBMS memory is needed with a row-level lock because the system must track a separate lock entry for every individual row

Why not others:
- (a) row / row — row-level lock causes less contention, not more
- (b) row / table — both parts reversed
- (d) table / table — table-level lock uses less memory (one lock per table)

Key rule: Finer granularity (row) = less contention but more memory overhead. Coarser granularity (table) = more contention but less memory overhead.

AI-generated — may contain errors