Data Modelling

[surajap123]

Hi Experts,

I have good understanding of dimensional modelling and know that we need to create either a star or snow flake schema depending on our requirement to build a dimensional model.

However I see in the Demo qlikview app, the data model doesnt follow the above rule. Having all the measures in a central fact table and attributes in surrounding dimension tables makes me comfortable to know what are measures and what are dimensions, so it becomes easy to select dimensions and create expressions(using measures) while building qlikview objects(eg-charts).

1.My question is if there is no need to follow the star schema model then, what is the easy process to build a good data model in qlikview?

2. I want to know if I am given access to OLTP Database to develop a qlikview app, Can I pull whatever tables I want into qlikview(as per the analysis requirement) and just removing the synthetic keys and stuff?(without worrying about creating a fact and dimension tables from that OLTP)

Your inputs are very valuable!!

Also point me to the resources pls..

Regards,

Suraj

[Josh_Good · Accepted solution]

Hi Suraj,

In QlikView you can often take shortcuts that would not be possible in traditional BI tools.  Your example, is a great one.  Technically it is 'better' to create a start schema to give faster performance however often the performance difference will not be noticeable so putting the effort into creating a star isn't worth it.  Many QlikView deployments uses a normalize data model.

Yes you can simply hit your OLTP database and use the tables in that structure.  Once you build your QlikView app if you find you need to move to a normalized structure then you can do it at that time without effecting your front end (assuming you keep the field names the same).  You likely won't have any issues unless you are getting into the 100 Millions of rows.

Regards,

Josh

[surajap123]

Hi Josh..your answer cleared my doubt..Thanks a ton !!