Relational Data Model

The relational data model represents all data as relations — sets of tuples over named attributes — and bases data manipulation on operations from set theory and first-order logic, achieving independence between the logical view of data and its physical storage.

Definition

In the relational model a relation is a set of tuples, each tuple a mapping from a fixed set of attributes to values drawn from those attributes' domains; a relational database is a collection of such relations governed by integrity constraints.

Scope

This topic covers the structural part of the relational model: relations, tuples, attributes and domains; schemas and instances; candidate, primary, and foreign keys; and the entity-integrity and referential-integrity constraints. It treats relations as having set semantics and a fixed arity, and explains how data independence frees applications from storage details. It excludes the query languages built on the model (covered under SQL and relational algebra) and the design discipline of normalization.

Core questions

• What are the structural components of a relation: attributes, domains, tuples, and arity?
• How do candidate, primary, and foreign keys identify and link tuples?
• What do entity integrity and referential integrity guarantee?
• How does the model achieve logical and physical data independence?
• Why are relations defined as sets rather than ordered lists of tuples?

Key concepts

• relation, tuple, attribute, domain
• relation schema and instance
• candidate and primary keys
• foreign keys
• entity integrity
• referential integrity
• logical and physical data independence
• null values

Key theories

Relations as sets of tuples

A relation is a subset of the Cartesian product of its attribute domains; because it is a set, tuples are unordered and duplicate-free, and the model inherits the algebra of set operations.

Data independence

By describing data purely in terms of relations and constraints, the model separates the logical schema from physical storage and access paths, so the physical representation can change without rewriting applications.

Integrity constraints

Entity integrity forbids null primary-key values and referential integrity requires every foreign-key value to match an existing primary key, ensuring that relationships among relations stay consistent.

Clinical relevance

The relational model is the basis of the dominant class of database systems used in business, finance, government, and the web; its clean separation of logical structure from physical storage is why relational databases have remained the default choice for structured, transactional data for half a century.

History

E. F. Codd proposed the relational model in 1970 at IBM, arguing against the prevailing hierarchical and network models whose queries were tied to physical navigation. Prototype systems System R (IBM) and Ingres (Berkeley) demonstrated its practicality in the mid-1970s, and Codd later received the Turing Award for the work.

Key figures

• Edgar F. Codd
• Jeffrey D. Ullman

Seminal works

• codd1970
• silberschatz2019

Frequently asked questions

Why are relations defined as sets, with no duplicate tuples?

Defining a relation as a set gives the model a clean mathematical foundation and means a tuple's identity is fully determined by its attribute values plus the key constraints. In practice SQL relaxes this to multiset (bag) semantics, allowing duplicate rows unless a key or DISTINCT enforces uniqueness.

What does referential integrity actually prevent?

It prevents dangling references: a foreign key in one relation must either be null or match an existing primary-key value in the referenced relation, so you cannot, for example, record an order for a customer who does not exist.