I'm having the luck (??) that i can access my data in 2 databases. The first one is a snowflake database with hundreds of tables containing all our transactional data. This database is synchronized with another database that has a star-structure. The second database is already running for years to create our SSAS cubes.
As we started with QV recently i'm abit wonderign what would be the 'better' and/or 'faster' approach. When using the snowflake i will reduce overhead of data-reduncancy, but the keys between the tables are textual-composed keys, and some of them are quite long.
The second database only use integers to reference other tables.
But, as for example the query to get our complete feature tree (we are a cinema company) involves 23 tables, in the star-version this is all in 1 table ....
If people have any experience on this choise, please feel free to comment.
In my experience, you won't see much difference in QlikView between star and snowflake layouts. It is perfectly happy with either, and performs just fine with either. Data redundancy isn't an issue in QlikView because you can't update and because its internal compression keeps it from requiring much more memory for redundant copies. I think I've actually seen memory usage go DOWN when denormalizing data in QlikView.
In your case, it sounds like the star version would be simpler for you to deal with, so that's probably what I'd recommend.
In my experience, you won't see much difference in QlikView between star and snowflake layouts. It is perfectly happy with either, and performs just fine with either. Data redundancy isn't an issue in QlikView because you can't update and because its internal compression keeps it from requiring much more memory for redundant copies. I think I've actually seen memory usage go DOWN when denormalizing data in QlikView.
In your case, it sounds like the star version would be simpler for you to deal with, so that's probably what I'd recommend.
