Previous Topic: PlanningNext Topic: Mapping Areas to Files


Segment Boundaries

Note the following section about segment boundaries.

One Schema One Segment

Typically one segment contains the data described by one schema. However, other factors need to be considered when deciding how data should be separated into segments.

Non-SQL Defined Data

Place all areas defined by a single schema in one segment unless:

If areas described by a single schema are separated into different segments, it is strongly recommended that no set crosses the segment boundary (no set should have an owner in one segment and a member in another). This is advisable because it will allow you to perform maintenance operations (such as reorganization) on a segment independently of other segments. If a set crosses a segment boundary, you may need to define a new segment that includes all impacted areas or create a database name that includes all impacted segments and whose name is the same as the segment on which the operation is being performed.

SQL Defined Data

Each table is associated directly with an area in which its data rows are stored. Restrictions about where the rows of a given table can be stored are imposed by security and the DBA when defining a segment.

A segment can be reserved for tables from a specific schema by specifying the FOR SCHEMA clause on the segment definition within the application dictionary in which the tables will be defined. By specifying the FOR SCHEMA clause, the DBA ensures that only tables associated with the named schema will be stored in the segment. This can be useful in ensuring that only related production data is stored in a given segment.

In an information center or development environment in which schemas are owned by individuals, it is likely that tables from multiple schemas will reside in a single segment. Segmentation might be related more to group affiliation than to schema association.