A database is a collection of data sets, data items, and records as shown in Figure D–1. Each box represents a part of the database and contains three labels. The first label, in uppercase letters, describes the part in conventional terms (for example, GROUP). The second label, in mixed capitalization, describes the part in Enterprise Database Server terms (for example, Data Set). The last label, in lowercase letters, describes the part in data management terms (for example, relation).
Each part can contain one or more of the parts underneath it. Thus, a database can contain one or more data sets.
The database arrangement can be created in two ways:
-
Bottom-up approach
The database is described from the lowest level (the record) up to the highest level (the database). The data items are collected in groups called data sets, and certain of these sets are designated as retrieval keys to optimize certain access paths. These data sets might or might not correspond to simple concepts in the real world.
Given a data set D, if an item or a collection of items in the data set (Y of D) uniquely determines another item or another collection of items (X of D), then X of D is functionally dependent on Y of D.
-
Top-down approach
The database is created from the highest level (the database) to the lowest (the record). This is generally a better approach, since people tend to organize data from the higher level to the lower one. With this approach, the designer begins with groups of data and successively refines the relationships between items in those groups. Thus, the data sets normally correspond to simple entities in the real world.
