How Are Tables Used?

The fact table and dimension tables are used in the star schema after you drag and drop them onto the workspace.

Because the star schema is formed in memory by the server, the first step is to make sure that the tables are loaded to memory. Be aware that the tables are transferred each time the star schema runs and this can reduce performance. Conversely, if the tables are already loaded to memory on the server, the star schema runs and generates the output extremely fast.

The first 15 characters of a dimension table’s name and the underscore character are initially set as a prefix for the column names from the dimension table. Column names for the output table are a combination of the prefix and the original column name.

However, you can specify a different value for the prefix after you select the table name from the menu at the top of the Inputs tab.

Column names from the fact table are not modified with a prefix.

To use memory efficiently, the default output for the schema is a view. The Create output as a view check box on the Output tab is selected by default to create a view. If you clear the check box, then the output is a table.

When the output is a view, the rows are formed in buffers rather than creating the entire table at once. This enables you to access a view whose data is larger than the memory capacity of the system.

If you clear the Create output as a view check box to create a table from the star schema, then the system must have enough available memory to store the table. If the system runs out of memory while running the star schema, the memory that was used for the output table is freed, and you receive an error message.

If you use SAS LASR Analytic Server tables for input to the schema, then the output table for the star schema must use the same library.