Before 1970, programs reached their data by following pointers through a specific storage layout, so changing the layout broke the programs that used it. Edgar Codd’s relational model swept that away with a simple, mathematical idea: store data as tables, and ask for what you want rather than spelling out how to fetch it.

The idea

Represent all data as relations, tables of rows and columns, and let users query declaratively, stating what information they want instead of the steps to retrieve it. The system figures out the how.

Codd’s model

Edgar F. Codd, working at IBM, set out the relational model in his landmark 1970 paper, having first described it in 1969. Data is represented as tuples grouped into relations, with attributes as columns, tuples as rows, and relations as tables. The whole thing rests on the mathematics of relations, which gives it a firm and simple foundation.

Declarative querying

The model provides a declarative method: users directly state what information the database contains and what they want from it, and the database management system works out how to retrieve it. SQL became the language that realized this approach, and decades later it still runs a very large share of the world’s data.

Why it matters

Separating what from how is one of the most productive ideas in all of computing. It let the physical storage change without rewriting the applications on top, and it founded the entire database industry. Underneath the tidy tables sits the messy reality of the file system and the disk, which the model deliberately hides.

Sources

  • “Relational model,” Wikipedia. https://en.wikipedia.org/wiki/Relational_model . Supports the model’s description by Edgar F. Codd (first described 1969, with his landmark 1970 paper), data represented as tuples grouped into relations with attributes as columns and tuples as rows, and the declarative method by which users state what information they want.