Before we can access and integrate data with ODI Designer, we need to create "models" based on the Logical Schema definitions we have in the topology. ODI models are abstracted to be independent of the physical data server type (database, file, JMS queues and topics, and so on), so they will have a common look and feel despite having dramatically different physical representations. This greatly simplifies their use as sources and targets and is a major benefit of using ODI.
Although we can manually create an ODI model, including its datastore definitions, constraints, and relationships, the normal and recommended method of creating models is by reverse-engineering the structural metadata directly from the databases, files, and other data resources. Sometimes a combination of both activities is used whereby the developer begins with the reverse-engineering process, then augments the captured metadata with additional metadata details (such as referential...
