Which Is Better For Performance - Suppress When Value is Null? OR Adding Set Analysis in Expression?

[benvatvandata]

Hi, 

I have an object that was having duplicating row issues for some Items... and I'm wondering which 'fix' is better for performance?

For example: 2 rows for 1 ItemCode... where one row had values for Product Group... and the other row was null...   

• If I add set analysis to one of my expressions, and edit one of my calculated dimensions... the problem is fixed... OR
• If I just click 'Suppress When Value Is Null' on the Product Group dimension... the problem is also fixed...

Which solution is better for performance?

Thanks,

Ben

[Brett_Bleess]

I would venture the best way to determine things is to set both and run a test monitoring the QV.exe process via Windows Task Manager and the Processes tab to track the QV.exe process to see how much CPU and memory is used by each method, that is the only scientific means I see to determine which one you should use in this case.  I would venture the Suppress Null is going to be the better option in this case, but I may be wrong.

Here is a Help link along optimization too, but it does not get into the area you are discussing unfortunately:

https://help.qlik.com/en-US/qlikview/April2020/Subsystems/Client/Content/QV_QlikView/application-per...

Regards,
Brett

[benvatvandata]

Hmm ok... I can try with this one object, although I'm thinking the difference may be extremely small to the point where it's not noticeable for just updating this one object (and with the memory varying slightly during the day I might not be able to tell which one is actually helping performance)...

was hoping to apply this logic to a bunch of objects after I confirmed which way was better (which then I might be able to actually see a difference in performance)... but for now I don't want to update all the other objects unless I know for sure which way is better (would take quite a while to update all of them).

Thank you for the reply though! 

- Ben

[MikeW]

In my experience, the set analysis should give better performance if you have a calculation that is complex enough to notice a difference. Set analysis will change the scope of the data your calculation is based on, but the suppress null still calculates your expressions for the null dimension. 

That said... if you are just trying to ignore problem rows in your data, then it's better to fix it in the data model.

[MyrtleFletcher]

hello everyone ,  In my experience updating this one object might not be able to tell which one is actually helpful but only scientific means I see to determine which one you should use is get some more reviews or try it  both. WalgreensListens

[Brett_Bleess]

Thought of one other post from Henric that might help, here is the YouTube recording of things, might shed some further light upon things for you too, it is Henric covering the QIX engine and how things are processed etc.

https://www.youtube.com/watch?v=wevhFK_AID8

Regards,
Brett