The Aggregate Designer consists of a number of components:

• The schema loader component ({@link org.pentaho.aggdes.algorithm.SchemaLoader}) reads a star schema and presents the algorithm with a {@link org.pentaho.aggdes.algorithm.Schema} object and embedded {@link org.pentaho.aggdes.algorithm.Table} and {@link org.pentaho.aggdes.algorithm.Attribute} objects.
• The algorithm component ({@link org.pentaho.aggdes.algorithm.Algorithm}) tries to find the most efficient set of aggregates for a given schema. A set of aggregates is efficient if it enables any expected query to evaluate quickly yet has a reasonable cost in terms of disk space and load time.
• The result handler component ({@link org.pentaho.aggdes.output.ResultHandler}) converts the result of the algorithm into a useful form such as a script of CREATE TABLE statements to create aggregate tables and INSERT statements to populate them. A result handler is not necessary if you are invoking the algorithm programmatically and your program will process the ({@link org.pentaho.aggdes.algorithm.Result})object produced by the algorithm itself.

Each of the components is specified a particular Java interface that extends {@link org.pentaho.aggdes.algorithm.Component}. There is a default implementation for each component, but you can provide your own.

There is a command-line utility {@link org.pentaho.aggdes.algorithm.Main} to allow you to invoke the algorithm. You can choose which implementation of each component, and provide parameters to each component.