Relational database components

At the highest level, relational databases have a lot in common with standard tabular organisation of data. Relational databases store data in tables; each row is an individual record and each record has the same collection of data points as the others, which are arranged in columns. So we have tables, rows and columns as the basis for a relational database, in common with data storage from old-school tabulation of astronomical data through to modern spreadsheets.

Relational databases differ from spreadsheets in how they are designed to store the data; linked tables are the norm and dependence of one data point on another is constrained by the normalization process (compare to spreadsheets' profusion of values calculated from other values in the same or another table). The underlying software of the RDBMS is optimised for preserving data consistency, with powerful querying via SQL, often with good support for multi-user access to the database, and predictable query performance.

Relational database designers have direct control over how tables are linked via specification of foreign keys; when enforced, these comprise constraints on what data record values can be stored in the database based on what other records are stored elsewhere. In addition, database designers and administrators can specify indexes on columns to improve SQL query performance.

All of these components are discussed in more detail in the following pages.