A few remarks regarding what I think you are trying to tell us:
- Usually there aren't any connections at all between different data sources, so they will not deliver any when you extract the data. You'll have to explicitly specify how the different data sources relate to each other. This may imply performing a good deal of transformations.
- In an RDBMS, connections are usually supplied by the queries your applications perform on the tables (and by the defintions of constraints and foreign keys). Extracting the data from an RDBMS doesn't include any of this connection information.
- QlikView connects fields and tables to one another by way of columns having the same name. But QlikView doesn't bother with the actual content of these columns, so the data may not match at all. You'll have to make sure that identically named columns actually contain the same type of data.
- Moreover, QlikView doesn't like the connection of two tables by way of multiple keys. Natural Joins à la QlikView have the habit of using multiple key fields, circular references and possibly a large amount of synthetic keys. All of them should be eliminated or replaced by something better.
When creating a QlikView data model, the existing data source architecture and query logic can be a big help for the design of an initial QlikView data model and the required associations. But there is only a small chance that you can import an existing data model as-is into QlikView and make it work without a change.
