Databases

Tables, rows, keys, and why the order of rows means nothing.

SELECT, WHERE, ORDER BY, LIMIT, and reading the error message instead of guessing.

INNER, LEFT, and the missing ON clause that quietly returns every pair of rows.

Why your COUNT is too high: join fan-out, GROUP BY grain, and DISTINCT as a symptom.

Three-valued logic, IS DISTINCT FROM, and why NOT IN with a NULL returns nothing at all.

Scalar and correlated subqueries, EXISTS as the safe rewrite, UNION and EXCEPT.

Entities, relationships, cardinality, and the join table every many-to-many needs.

First through third normal form, and the update anomaly each one removes.

text, numeric, timestamptz, enums, and the types Postgres itself tells you to avoid.

Identity columns, uuidv7, NOT NULL, CHECK, foreign keys, and what NOT ENFORCED is for.

Versioned schema changes, reversibility, and the ALTER TABLE that locks the table for an hour.

How a row is laid out on disk and what happens to values too big to fit in a page.

BEGIN, COMMIT, and what read committed, repeatable read, and serializable each guarantee.

Row and table locks, the queue behind them, and reading pg_locks when everything stops.

Row versions, dead tuples, bloat, autovacuum, and freezing before the wraparound warning.

B-tree lookups end to end, and where GIN, GiST, and BRIN beat a B-tree.

EXPLAIN ANALYZE, estimated versus actual rows, and the buffer counts now printed by default.

Column order, composite, partial, covering and expression indexes, and skip scan in Postgres 18.

ANALYZE, why the planner guessed wrong, pg_stat_statements, work_mem, and connection pooling.

Logical dumps, physical base backups, WAL archiving, and restoring to a chosen timestamp.

Streaming replicas, replication lag, read routing, and what a failover actually loses.

Declarative range partitioning, and a major version upgrade that keeps its optimizer statistics.

jsonb, full-text search, pgvector similarity search, and when a separate store is worth the cost.