Relational database
Data held in tables linked by keys and queried in a standard language.
Computing and AI · Software · No. 473
- Digital age
- Readiness 9 of 9
- Platform
A 1970 paper by an IBM mathematician still decides how banks, travel firms and phones store their records.
Chart: Illustration: Relational database
Where it comes from
Chart: Map of where it comes from
Edgar Codd, a mathematician at IBM's San Jose lab, proposed in 1970 that data be kept in linked tables. IBM tested the idea in its System R project from 1973.
Three things to know
- Codd's 1970 model let users get a new table from one or more tables with a single query, without knowing where data was stored.
- IBM's Chamberlin and Boyce described the query language SEQUEL, later SQL, in 1974, and ISO made SQL a standard in 1986.
- Oracle shipped its first product in 1979 and by 2000 was the world's second biggest software company.
The journey
- 1 ModelFacts are split into tables, one row per record and one column per attribute
- 2 KeysShared values such as a customer number link rows in different tables
- 3 QueryA user states in SQL what data is wanted, not how to find it
- 4 OptimiserThe system works out a fast plan using indexes and joins
- 5 TransactionChanges are written so that each one completes fully or not at all