Relational Database Model Principles and Evolution

2

The Relational Model

Relational Model: Defines a database in terms of objects, properties, and operations. Objects with the same structure and behavior belong to the same class, and classes are organized into hierarchies. Relational models have extended their models to incorporate object-oriented concepts. In 1970, Edward F. Codd published an article in which he argued that data should be inter-linked by natural and logical relationships inherent to the data, rather than through physical pointers.

Codd proposed a simple data model in which all data are represented in tables consisting of rows and columns. Codd also proposed two languages to manipulate data in tables:

• Relational algebra
• Relational calculus

The logical manipulation of data also makes the creation of query languages more accessible to users who are not programming specialists. Relational systems are now a standard in the market, especially for commercial transactions. Examples of relational database systems include:

• Informix
• DB2
• Oracle
• Sybase
• DBASE
• FOXPRO
• INGRES
• MySQL
• MS SQL Server

Key Features of the Relational Data Model

The relational model is more efficient than the hierarchical and network models. A database is represented as a collection of tables (relations) that can be stored individually in file form. Any line is determined in a relation called a tuple, and each column represents a different domain relationship.

Differences from Hierarchical and Network Models

The relational model is completely different from the hierarchical and network models, not only in its architecture but also in the following points:

• Independence in implementation: It is not necessary to know how data are physically implemented (no need to work with pointers, linked lists, graphs, etc.).
• Terminology: The relational model has its own specific terminology.
• Logical keys as pointers: Use primary and foreign keys to represent the associations between two files. However, due to independence from implementation, the system can use physical database pointers or other methods, but these are transparent to the user.
• Normalization theory: This theory was developed in the context of the relational model, but today its properties have been extended to other models. It is a set of properties that data must meet to achieve a database design free of anomalies and with minimum redundancy.
• Comprehensive programming languages: There are simple languages to access the relational database. These are languages that can manipulate data as groups or files, instead of one record at a time like traditional procedural languages.

Historical Development of the Relational Model

Summary of the historical development of the relational model (1970-1990):

1970-1978: Development of the Theoretical Model

• Definition of the model (Codd 1970-72)
• Relational algebra and calculus
• First design problems

1975-1980: First Relational DBMS Development

• IBM (System R)
• SQL (1975)
• DB2, SQL/DS, OS/2, SQL/200
• INGRES Project

1980-1990: Large Relational Systems

Development of large-scale relational systems.