Topic 8 · Databases (HL Option)

Row = record/tuple. Column = attribute/field. Each row UNIQUE by primary key.

Column(s) uniquely identifying each row. Often an auto-increment integer.

Column referencing the PK of ANOTHER table. Links tables. Enforces REFERENTIAL INTEGRITY.

Combine rows from 2+ tables on matching key. INNER (only matches), LEFT (all left + matches), etc.

Atomic values only (no lists in cells); each row unique.

1NF + every non-key column depends on the FULL primary key.

2NF + non-key columns depend ONLY on PK, not on other non-key columns.

Eliminate redundancy → no UPDATE/INSERT/DELETE anomalies → data consistency.

Reverse for warehouses: duplicate columns to make analytical queries fast.

`SELECT cols FROM table WHERE cond GROUP BY g HAVING agg-cond ORDER BY col LIMIT n`.

Atomicity (all or nothing) · Consistency (valid state) · Isolation (concurrent = serial) · Durability (survives crash).

Group of operations that succeed/fail TOGETHER. BEGIN/COMMIT/ROLLBACK.

Confusing PK (uniqueness) with FK (foreign reference). Both are 'keys' but very different roles.