Overview

TIBCO Patterns supports the concept of joins. Using joins you can search data spread across multiple tables in a way that the fields of each table appear to be merged into a single table.

Before the joins feature was introduced, there were two basic methods for performing a join query. The first is to merge the records of the tables before loading them into a TIBCO Patterns server using a denormalization process. The denormalization process involves creating a separate merged record for each valid combination of records from the individual tables. Denormalization is practical when only two tables are involved. For more than two tables, denormalization can result in creating very large tables. An update operation on a record in one of the separate tables requires updating many records in the merged table, making it difficult to maintain the denormalized table.

The second method is to query each table separately and then merge the query results. With this approach, you might lose accuracy. The result might not include the best merged record. The best merged record might contain one or more individual records that match very poorly. If each table is queried individually, these poorly matching records are not returned. Hence, the best merged record is lost.

Using the joined tables feature of TIBCO Patterns to do a joined search eliminates the issues with denormalization and merging of separate query results.