Skip to main content

Module Quiz

Complete this quiz after finishing all concept and practice pages. Save your answers before opening separate quiz calibration.

Current Module Questions​

Question 1: ACID Property Mapping​

Which of A, C, I, D is most directly supported by each mechanism: (a) the write-ahead log; (b) a foreign-key constraint; (c) row-level write locks; (d) an fsync on COMMIT; (e) a CHECK constraint?

Question 2: ARIES Recovery​

During recovery from this log, what does ARIES redo and undo?

10 T1 BEGIN
11 T1 UPDATE x: 5 -> 7
12 CHECKPOINT (active: T1; data file x=5)
13 T2 BEGIN
14 T2 UPDATE y: 2 -> 4
15 T1 COMMIT
16 T3 BEGIN
17 T3 UPDATE z: 8 -> 9
   -- crash --

Question 3: Anomaly Identification​

A transaction T_total reads account A (balance 500), then account B (balance 200), and returns the sum. Concurrently, T_xfer transfers 100 from A to B. T_total returns 600 when the true sum should be 700. Name the anomaly and the weakest isolation level that prevents it.

Question 4: Anomaly Identification​

Two concurrent transactions each execute:

SELECT COUNT(*) FROM doctors WHERE on_call;  -- both return 2
UPDATE doctors SET on_call = FALSE WHERE name = 'self';

Both commit under PostgreSQL REPEATABLE READ. The "at least one on-call" invariant is now violated. Name the anomaly and the weakest prevention.

Question 5: Isolation-Level Choice​

A workflow reads a user's current profile, validates a change, and writes the new version. Two admins sometimes edit the same profile concurrently. Choose the weakest PostgreSQL isolation level that prevents lost updates without over-serializing, and name an alternative primitive.

Question 6: Anomaly Identification​

An inventory system sold the last item twice because two checkouts both read quantity = 1, both decremented, and both committed. Name the anomaly and give two distinct preventions.

Question 7: Anomaly Identification​

Two users sign up simultaneously with the same email. Each transaction ran SELECT 1 FROM users WHERE email = ? (no row), then INSERT INTO users (email, ...). Both succeeded under READ COMMITTED. Name the anomaly and two preventions.

Question 8: Isolation-Level Choice​

A dashboard issues a read-only query computing totals, revenue, and counts across many tables. Latency matters; stale-by-a-few-seconds is fine. Choose the right isolation/read strategy on PostgreSQL.

Question 9: 2PL vs SI Pick​

For a workload of 80% short writes, 20% long reads, which concurrency-control scheme reduces contention: 2PL or SI/MVCC? Why?

Question 10: 2PC Failure Mode​

In 2PC, a participant has replied PREPARED and the coordinator then crashes before sending COMMIT or ABORT. What must the participant do, and why can it not unilaterally abort?

Question 11: Saga Design​

For a travel-booking workflow (reserve flight, reserve hotel, charge card), write the compensations for each step and identify any that resist compensation.

Question 12: Linearizability Analysis​

Is this history linearizable? C1: write(x, 1) (t=1-2); C2: read(x) -> 0 (t=3-4); C3: read(x) -> 1 (t=5-6); C4: read(x) -> 0 (t=7-8).

Question 13: Causal vs Eventual​

Give a concrete anomaly that causal consistency prevents but eventual consistency allows.

Question 14: CAP and PACELC​

A vendor says "we provide strong consistency and high availability." Apply CAP and PACELC honestly. What question do you ask to force clarity?

Question 15: Isolation-Level Choice​

An audit job reads the entire transactions table once and writes an aggregated report to another table. Concurrent writers are constantly inserting new transactions. What isolation level produces a consistent snapshot without aborting the audit, and what is the cost?

Interleaved Review Questions​

Prior Module Question 1 (S6 Module 3: Replication)​

What is the difference between synchronous and asynchronous replication, and which ACID property is most affected by the choice?

Prior Module Question 2 (S6 Module 2: Storage Engines)​

Why is a log-structured merge-tree (LSM) write-optimized compared to a B-tree?

Prior Module Question 3 (S6 Module 1: SQL)​

Explain why declarative UPDATE counter SET n = n + 1 WHERE id = $id is usually better than SELECT n ... / UPDATE n = $computed for concurrency.

Prior Module Question 4 (S5 Module 3: OS memory)​

Why does an fsync on a database commit not fully guarantee durability in the presence of a hardware write cache?

Prior Module Question 5 (S5 Module 4: Networks)​

In a data-center network, why is "message delay is unbounded" a more accurate assumption than "message delay is bounded with high probability"?

Self-Assessment and Remediation​

Mastery Level (90-100% correct):

  • Ready for Module 5 (distributed systems fundamentals) with confidence.

Proficient Level (75-89% correct):

  • Review missed concepts, redo one practice lab for each missed area.

Developing Level (60-74% correct):

  • Redo Practice 1 (anomalies lab) end to end. The gap is almost always in the anomaly-recognition reflex.

Insufficient Level (<60% correct):

  • Restart with Cluster 2 (anomalies). Most likely issue: pattern-matching isolation levels without simulating the interleaving. Work every anomaly by hand on paper before touching code.

Progression boundary​

Quiz performance calibrates retrieval and reasoning. It does not replace runnable work, changed-case transfer, the module assessment or required semester evidence. Consult practice calibration for the additional independent cases.