What is a Relational Database?

Note: Click the (plus icon) to show more details. Click the (minus icon) to hide details.

---

• Where Did Relational Databases Come From?
• Primary Attributes of a Relational Database
• Some Relational Operators

Where Did Relational Databases Come From?

• The concept of Relational Databases was formulated in 1969 by E.F. Codd of IBM
• The term "relational" comes from the mathematical studies of "relational algebra" and "relational calculus" that were used to manipulate this type of database model. Fortunately, we don't have to know algebra nor calculus to use a relational database.
• Codd specified 333 criteria for a relational database. Of these, 12 are considered the core criteria. We'll discuss a handful of these core criteria in a minute.
• As of this writing, there are NO database products that qualify fully as a relational database (i.e., that meet Codd's criteria).

Top of page  

Primary Attributes of a Relational Database

• A relational database is a collection of two dimensional tables. Think of the horizontal dimension as rows and the vertical dimension as columns. You can also think of them as records and fields. Each table has exactly four sides: that is, each row in a table has the exact same number of columns (i.e., every record in a table has the exact same fields).
• Each table's attributes (more commonly known as columns of fields) must all be of the same data type. For example, in our scenario, each dog can be uniquely identified by a dog number. Therefore, our table should have a field/column for tracking the dog number, which can only contain numbers. We can't mix a numeric dog number with alphabetic dog codes. This can be a confusing concept at first. For example, a social security number, while it has digits is really a code. But a price is always a numeric value. You can't have a relational database where the price is stored as 89.95 in one record, but "a whole lot" in another. They must be the same data type.
• No two tuples (more commonly known as rows or records) are identical. Why would you want duplicate rows anyway? That would be a waste. Besides, when we normalize a relational database, one step is to eliminate wasteful duplication.
• The order of the rows is not important. This rule allows us to use indices to sort the data in multiple ways. With a relational database we can insert new rows at the beginning, in the middle, or at the end. Also, when data is deleted in a relational database, it doesn't require any kind of reorganization.
• Each attribute must have a domain. This rule states that each column has specified limits. For example, a dog number might have a limit of three digits. It's domain would be any integer between zero and 999.

Top of page  

Some Relational Operators

Do you remember "sets" from elementary school? Tables in a relational database are data sets, so you can perform relational operations on them. Here are a few of the relational operators which can be performed on a relational database:

	UNION: data sets combined.
	INTERSECTION: only the data shared between sets.
	DIFFERENCE: only data not shared with another set.
	SYMMETRIC DIFFERENCE: only data not shared with all sets.

Top of page