Data model for QlikView and Impact of NULL

[blade001]

Hi,

I just wanted a second opinion on a QlikView data model I have been asked to optimize. The model it self seems rather overly complicated and doesn't follow any conventional dimensional modeling.

I am currently looking at the the model and trying to reduce number of tables it contains to remove having too many joins and fall back to a slightly larger (lots of columns) dimension tables rather than having 4-5 tables associated with each dimensions (going more towards the star schema from snowflake). Please see the attached diagram below. I prefer the model 2 as this will simplify the model. The current model looks like the 1st model demonstrated in my diagram. Now I can't see any advantage of having the data model 1 over the data model 2. Also Just wondering having the NULLs in the Model 2 is a bad thing when scaling the data model and what impact each model will have on memory? Any thought would be more than welcome.

S.K.Rab

[awhitfield]

GO for option 2, keep it as simple as possible

Andy

[#3]

your results should be the same regardless of the model you pick, but I too, would go with 2.  It is simple, straight forward and puts all of the data into 1 (fact) table.

[#4]

Model approach gave better performance on UI but it make some additional refresh time on the large data sets.