Applications Development Appendix

Flat-file

Relational

Data normalization

Flat-file databases

In a flat-file database system only one table is used at a time. For small and simple applications such as a phone book a flat-file system is more than adequate. One problem with flat-file systems is a tendancy towards redundant information. For example, in a Sales Order system the customer's address would be repeated in every order placed by that customer. Apart from the increased data entry work this entails problems may arise where a customer changes address; every delivery note for each order not completed before the customer moves would have to have the address changed individually. A better solution would be to use a relational database.

Relational databases

Relational databases can maintain different information in different table and combine information from several tables as required. This makes relational databases more powerful and more complex than flat-file databases. By using more than one table repetition of data can be avoided making the database more efficient and more flexible. The process of reducing repetition is referred to as Normalization.

Data normalization

A few large complex tables are usually note the most efficient way to organize information in a database. Data normalization distributed information over many tables but with fewer fields.

In a fully normalized database only thoses fields that link tables would be duplicated. Although normalization is usually desirable to increase efficiency there are cases where a certain amount of de-normalization can make operations quicker.