Database TransactionsLearn from verified, runnable examples

Isolation levels, anomalies, locking, MVCC, and concurrency patterns, with asserted database runs, documented contracts, and marked derivations.

Start here

How this site works

illustrative deadlock, PostgreSQL

A> BEGIN;
B> BEGIN;
A> UPDATE accounts SET balance = balance - 100 WHERE id = 1;
B> UPDATE accounts SET balance = balance - 50 WHERE id = 2;
A> UPDATE accounts SET balance = balance + 100 WHERE id = 2;
⏳ A  waiting on B's lock (row id=2)
B> UPDATE accounts SET balance = balance + 50 WHERE id = 1;
⏳ B  waiting on A's lock (row id=1)
⏵ B  ERROR 40P01: deadlock detected
A> COMMIT;
⏵ A  COMMIT, shown survivor

✅

Verified, not vibed

Scenario transcripts come from real PostgreSQL or MySQL runs. Database-relevant CI changes replay the schedules and reject artifact drift. Manual contracts and marked derivations support broader explanations.

🧪

Run it yourself

`docker compose up -d --wait`, then `bun test`. Replay the Scenarios and change their schedules or isolation levels to test your predictions. Some reference topics use manual support rather than a Scenario.

🧠

Focused on what breaks

Study lost updates, deadlocks, phantom reads, and repairs with their transaction and writer boundaries stated. These demonstrations do not measure production workload performance.

The curriculum

Eight chapters per engine, with asserted examples and scoped reference material. Pick a database and start anywhere.

Transactions 101PostgreSQL MySQL
Isolation levels & anomaliesPostgreSQL MySQL
LockingPostgreSQL MySQL
MVCC under the hoodPostgreSQL MySQL
Real-world patternsPostgreSQL MySQL
Transactions across servicesPostgreSQL MySQL
Pitfalls compendiumPostgreSQL MySQL
ProductionPostgreSQL MySQL