Relational databases were first described by Dr. Edgar Codd during the process of his research in database management systems for IBM. Relational databases provide a table-oriented view of data: that is, data is perceived as a set of rows and columns. For example, in a table that contains information about a company's employees, each column in the table describes a different attribute of an employee (last name, home address, etc.), and each row in the table describes a different employee (John Smith, Jane Doe, etc.).

Table 1. A Table in a Relational Database

Crucial to the design of relational databases is the definition of the relationships among tables in the databases: for example, when a database designer defines an employee table and a manager table, the designer also defines the relationships between the two tables. The importance of the relationships among database tables is the reason for the word "relational" in the name "Relational Database Management System."

The major characteristics of an RDBMS are:

Declarative Manipulation: There is a consistent set of operators that manipulate the data. The data manipulation language is declarative, not procedural: for example, programmers specify what data they want to work with, but not the process by which the database should compute the data. Declarative manipulation is implemented in the SQL language, which we discuss later in this lesson.

Reducing Redundancy: Data redundancy is kept to a minimum using a technique called normalization, thereby simplifying the structure of the data in the database, as well as the overall size of the database.

Simplicity: There are a limited number of data object types and operators, as well as simplicity in presenting data because it is in a table.

Data Independence: The structure of the data in the database is independent of the user interface and business logic.