Database Management Systems

The relational model represents data as relations (tables) of tuples over typed attributes; keys identify tuples uniquely and foreign keys link relations, while integrity constraints keep the data consistent.

SQL retrieves data declaratively; joins combine rows from multiple tables on matching columns, and GROUP BY with aggregate functions summarises groups, with WHERE filtering rows before grouping and HAVING filtering groups after.

Normalization decomposes tables to remove redundancy and update anomalies, guided by functional dependencies; each normal form (1NF→2NF→3NF→BCNF) eliminates a specific kind of problematic dependency.

A transaction is an all-or-nothing unit that preserves ACID properties; concurrency control (such as two-phase locking) ensures interleaved transactions produce a serialisable schedule equivalent to some serial order.

Indexes speed lookups by maintaining an ordered or hashed access path — B+ trees for range and point queries, hash indexes for equality — while the query optimiser uses table statistics to choose the cheapest execution plan.