How Data Is Stored

A gentle, intuition-first look at how a database actually keeps information — rows, columns, and tables — without any scary internals.

You do not need to know how a car engine works to drive. Likewise, you do not need to know the deep internals of a database to use one well. But a little intuition about how data is stored will make everything that follows click into place.

From facts to rows

Almost all useful data is a collection of facts about things. "This customer is named Ada and lives in Toronto." "This order was placed on March 3rd for $42." A database's job is to keep millions of such facts tidy.

The relational database's big idea is to group facts into neat rectangles. Each rectangle is a table. Inside a table:

A row is one thing — one customer, one order, one product.
A column is one attribute of those things — a name, a city, a price.
A cell (one row, one column) holds a single value.

That picture — a labeled grid of rows and columns — is the single most important image in this entire course. Almost everything we do is either reading such a grid, changing it, or combining two of them.

Why a grid?

Why store data as rectangles instead of, say, paragraphs of text? Because a grid makes questions answerable in a uniform way:

"How many rows are there?" → count the rows.
"Which rows have city = 'Toronto'?" → check one column.
"What is the average price?" → look down the price column.

Every value has a known place (which row, which column) and a known kind (text, number, date). That regularity is what lets a computer answer questions about the data mechanically and quickly.

Same shape, everywhere

This rows-and-columns shape is the same one you see in a spreadsheet — and that is no accident. The grid is just an extremely natural way to lay out facts about many similar things. The difference is what the system around the grid can do.

A table is described by its columns

Before a table can hold any rows, the database needs to know its shape: what columns exist and what kind of value each holds. This description is called the table's schema.

You define the schema once ("a customer has a numeric id, a text name, and a text city"), and from then on every row must fit that shape. This is a feature, not a restriction: because every row has the same columns of the same kinds, the database can store, check, and search them efficiently and catch mistakes early.

Seeing it for real

Here is the table from the diagram, brought to life. Click Run to see the stored grid come back to you.

The * means "every column." The result you see — three rows, three columns — is the stored data. We will spend the whole Querying Data section learning to ask for exactly the slices we want instead of everything.

Check your understanding

QuestionSelect one

In a relational table, what does a single row represent?

One attribute, such as "city," shared by all the data.

One thing — for example one customer or one order — with a value for each column.

The name of the table.

The rules describing what columns the table has.

QuestionSelect one

A table's schema describes what about the table?

The current number of rows it contains.

Its structure — which columns exist and what kind of value each column holds.

The colors used to display the table.

The list of people allowed to read the table.