In a normalized database, there are typically master tables and transactional tables. The master table data is shared by many transactional tables, or a given transactional table may have multiple columns as foreign keys for the same master table. In a query, this type of transactional table(s) are joined to the same master table, the master table is repeated in the SQL query, for example,
SELECT <Required_Column> FROM YFS_ORGANIZATION O, YFS_PERSON_INFO PI1,
YFS_PERSON_INFO PI2,
YFS_PERSON_INFO PI3
WHERE O.CORPORATE_ADDRESS_KEY = PI1.PERSON_INFO_KEY (+) AND
O.CONTACT_ADDRESS_KEY = PI2.PERSON_INFO_KEY (+) AND O.BILLING_ADDRESS_KEY =
PI3.PERSON_INFO_KEY (+)
In this example, the Person Info master table has been repeated for each referencing column. If the Organization table was joined to another table that had more references to the Person Info table, you could have viewed see more Person Info tables in the query.
In the Data Model, joins are created between the entities to represent the relationships between them. Naturally, if there is only one entity for a master table, and more than one reference to it is required in a particular query, the SQL that is produced will not be correct, for example,
SELECT <Required_Column> FROM YNA_ORDER_LINE_VW OL,
YNA_ORGANIZATION_SHIP_NODES_VW SN
WHERE OL.SHIP_NODE_KEY = SN.SHIP_NODE_KEY AND OL.RECEIVING_NODE =
SN.SHIP_NODE_KEY .
This query is not correct. Therefore, in this case, you need two entities in the Catalog that correspond to the YNA_ORGANIZATION_SHIP_NODES_VW view. This is done by replicating the entity Organization Distribution Nodes into Organization Receiving Nodes and Organization Ship Nodes and separate joins are created with the Order Line entity.
If more entities require the same set of master tables, the master tables can be further replicated in the Catalog, for example, Organization Receiving Nodes and Organization Ship Nodes are required both by Order Line and Order Release. Therefore, there are two instances of each Node table in the Catalog.
Other tables may have to be similarly replicated from views. For details on these views, see the topic, "IBM® Sterling Warehouse Management System: Business Intelligence Views". Because this replication is extended for more and more master tables, the Star model of the database begins to look like a Snowflake. The following figure illustrates this concept.