Re: Why is a Concatenated table bad for the RAM consumption compared to Star Schema?
You need to clarify your terminology - what do you call a "concatenated table" and how is it different, in your opinion, from a Star Schema.
The terminology that I use is the following:
Star Schema is a kind of an analytical data structure (as opposed to a transactional one) that satisfied the following Requirements:
1. There is only one Fact table that contains all the transactions in the data model.
2. There is a number of Dimensional tables surrounding the Fact, and each Dimensional table is located not further than 1 link away from the Fact (in other words, all Dimensional tables are associated to the Fact but not to other Dimensional tables).
3. All the fields that participated in Expressions (metrics, flags, etc...) are stored in the Fact table.
Concatenated table is one of the commonly used data modelling techniques that is used to combine multiple transactional tables into a single ("concatenated") Fact table. The result of this technique is usually a Star Schema, or a Snowflake if Dimensional tables happen to be associated between them.
Star Schema is typically the best performing option of all, for two main reasons - the metrics are all stored in a single table, and Dimensions are not further than one link away. That being said, a Snowflake schema may only be 10-15% slower than a Star. This difference can be negligible in smaller data sets.
Now, having clarified the terminology, would you like to clarify your question?
You can read more about QlikView data modeling in my new book "QlikView Your Business" that just become available for purchase.