The relational database
A way of organizing data as simple tables — relations — of rows and columns, linked to one another only by the values they share, not by pointers or a fixed hierarchy. You ask for data by describing what you want, in a language built on set theory and logic, and let the system work out how to fetch it.
✦ Wait, really?
IBM had the blueprint first: E. F. Codd published the relational model from IBM's own San Jose research lab in 1970, and IBM built a working prototype, System R, complete with the query language that became SQL. But the company dragged its feet to protect an older database product it was already selling. So an outsider who simply read the published research, Larry Ellison, shipped the first commercial SQL database — Oracle, in 1979 — two years before IBM shipped its own. A mathematician's paper built one of the great software fortunes on Earth, for someone else.
What it is
A relational database stores data as tables. Each table is a relation: a set of rows, all with the same named columns, where a row is one fact — one customer, one order, one transaction. Tables are connected not by embedded links but by shared values: an order row carries the id of its customer, and that shared id is the only thing joining the two tables. Crucially, you never tell the system *how* to walk the data. You state *what* you want — "the names of customers in Berlin who ordered in June" — in a query language, and the system's optimizer decides how to find it. Codd's radical move was to separate the logical shape of the data, which people reason about, from the physical storage, which the machine manages.
Why it mattered
The databases before Codd's were navigational: data was strung together with explicit pointers into hierarchies or networks, and to retrieve anything a programmer had to know the physical layout and hand-code a path through it. Change the layout and every program broke. Codd, trained as a mathematician, argued that this was backwards — that data should be described by its logical relationships and queried by a language grounded in logic, with the storage details hidden. That gave three things at once: a rigorous foundation (query results are provably just set operations), independence (you can reorganize the storage without rewriting the queries), and accessibility (a non-programmer can ask questions in something close to structured English). It turned data management from a craft of pointer-chasing into a branch of applied logic.
What it unlocked
The relational database became the default way institutions store their records — banks, airlines, governments, retailers, and nearly every website with an account or a shopping cart runs on one. SQL, born in System R, is now among the most widely spoken computer languages on Earth, understood by dozens of competing database engines. The separation Codd insisted on — logical model above, physical storage below — let hardware and storage evolve for fifty years without breaking the applications on top, and even the newer "NoSQL" systems define themselves in relation to the model he wrote down in a single 1970 paper.
Minimum viable version
Two tables and a shared key: a table of customers and a table of orders, linked only because each order stores a customer's id. To find a customer's orders you match values, not follow a pointer — and a query language works out the matching for you.
This entry is awaiting its full account — the cartographers are at work. Its place in the graph is already verified.
Unlocked
Frontier — nothing charted yet.
Sources
- — E. F. Codd, 'A Relational Model of Data for Large Shared Data Banks,' *Communications of the ACM* 13, no. 6 (June 1970): 377–387
- — Donald D. Chamberlin & Raymond F. Boyce, 'SEQUEL: A Structured English Query Language' (1974)
- — C. J. Date, *An Introduction to Database Systems* (8th ed., 2003)
Something wrong on this page? Every claim here is meant to survive challenge. Suggest a correction →