What Is a Relational Database?

What Is a Relational Database?

What is a database, anyway?

A database is a system that gathers data in one place and gives it back when asked.

It can hold organized information about people, products, orders, and more (Oracle, “What Is a Database?”).

Unlike a file at hand, many people and apps can use the same data.

A relational database is a TV station’s archive

TV sometimes runs highlight reels of past footage, often before a big sports event.

From decades of footage, just the needed scenes come together in one program.

That works because the station’s archive is well organized, and it has an editor.

TV station

Director
     |  asks    ^  hands over
     v          |
Editor
     |  looks   ^  takes out
     v          |
Archive (footage)

A director asks, “I only want this athlete’s scenes from the finals.”

The editor finds them in the archive and hands them over ready for the program.

A relational database works in much the same way.

The editor of a relational database is called a DBMS (database management system).

Relational database

Director (person, app)
     |  asks    ^  hands over
     v          |
Editor (DBMS)
     |  looks   ^  takes out
     v          |
Archive (data in tables)

Split into tables, linked by number

In the archive, each clip gets a number and goes into a list.

The list of clips and the list of athletes are separate tables, linked by number.

Clips
Clip  Content    Athlete
5001  Final      101
5002  Semifinal  102
5003  Qualifier  101

Athletes
Athlete  Name
101      Sato
102      Suzuki

Clip 5001 has athlete number 101. Look up 101 in the athlete list, and you get Sato.

If the spelling of an athlete’s name changes, one fix in the athlete list is enough.

Relational means “related to each other.”

Each single table is called a database table.

Hold data in tables and link the tables by number. That is a relational database.

How it differs from Excel

Excel is closer to home videos lined up on your own shelf.

You shoot them, you arrange them, and you dig through them when you want one.

Spreadsheets were originally designed for one user (Oracle, “What Is a Database?”).

Nobody stops you from writing anything, but keeping things tidy and safe is up to you.

	Excel	Relational database
Where it lives	A file at hand	Gathered in one place
Who uses it	Usually one person	Many people and apps at once
How you write	Freely	Only in the set form
How you search	You look yourself	Ask, and only matches come back

Excel is not worse. It suits a different situation.

Excel for one person, a relational database for everyone.

Why it holds up when everyone uses it

Unlike an Excel file, it is built for many users from the start.

• Asked at the same time: it handles connections from several computers at once (PostgreSQL, “Architectural Fundamentals”).
• Stopped halfway: it bundles several steps into one that happens completely or not at all (PostgreSQL, “Transactions”). No half-finished reel goes out.
• Easy to back up: with everything in one place, one backup covers it. That also makes regular backups essential (PostgreSQL, “Backup and Restore”).

That all-or-nothing bundle is called a transaction.

Say a job is being reassigned from Sato to Suzuki.

If it stops right after taking the hours from Sato, the totals no longer match.

With a transaction, a failure means none of it happened.

Takeaways

A relational database stores things, of course.

But for the one job of finding them, it is remarkably effective.

Footage you cannot find efficiently might as well not exist.

With a system that links things so you can find them, you can make your own best-of reel anytime!

Trivia

• Database: A general name for systems that store data and give it back.
• DBMS: Software that handles a database. Short for database management system.
• PostgreSQL: Free relational database software.
• Transaction: A set of steps that happens completely or not at all.
• IDS: An early DBMS born at GE in the 1960s.

History

The beginning: born from factory production control

In the 1960s, the amount of data computers handled grew quickly (IBM, “The relational database”).

Around this time, Charles Bachman at General Electric (GE) built a system for controlling factory production.

From it came the Integrated Data Store (IDS), described as the first disk-based DBMS used in everyday production (Computer History Museum, “Charlie Bachman”).

Bachman received the Turing Award for this work in 1973 (IEEE Computer Society, “Charles Bachman”).

Walls and turning points: from paths to tables

But databases of that era were tied to parent-child links and storage paths (IBM, “The relational database”).

In 1970, Edgar Codd of IBM published the idea of holding data in tables (Codd, “A Relational Model of Data for Large Shared Data Banks”).

Linking data by the values inside it, not by where it was stored, was the new part.

IBM tested it with System R from 1973 and shipped the DB2 product in 1983 (IBM, “The relational database”).

This form spread in the 1980s and is still the mainstream (Oracle, “What Is a Database?”).

The next wall was the internet.

In 2007, Amazon described Dynamo, a system that keeps accepting shopping cart writes even during failures (Werner Vogels, “Amazon’s Dynamo”).

It gave up some consistency in order to keep running.

Tables stopped being the only answer, and the choice came to depend on the use.

What is in development now

• In May 2025, AWS made Amazon Aurora DSQL generally available, a distributed database built for high availability with no servers to manage (AWS, May 27, 2025).
• In October 2026, the first release candidate of PostgreSQL 19 came out (PostgreSQL, October 15, 2026).

What comes next

Archives have spread from in-house servers to outside services.

Even so, holding data in tables and linking them by number remains the idea from 1970.

Whether tables last another 50 years is still unknown.