Best Praxis/Performance

[Anonymous]

I have setup some Alternate State, so a lot of my charts, will which forecast it should compare with.

For this purpose i wan't to change all labels and expressions.  And it got me thinking, if there could be a performance advantage with using a variable?

Today we have a label which could say  "Var F1"   Which could be replaced with ='Var '&Concat({[Primary Forecast]}[Forecast Actualities])   which then shows the choosen actuality in a selection box.   but this label will maybe occur 200 times within my qlik application.    Would i have more efficiency by creating a variable called  VARACT  which have the formula  ='Var '&Concat({[Primary Forecast]}[Forecast Actualities])   So all 200 times would look into the variable?

I can see the big advantage, if i ever choose to change the text from VAR to DIFF  or similar.  but are there are performance effect? And what is best practice, if you think on streamlining and overview for future people who needs to do changes? 

My thought were that 200 labels looking at one variable(which has a formula) are lighter in performance than, 200 labels doing the same formula.  This is only 1 example, it would be replicated with sales, gross margin etc. 

[luke1986]

Hi Mike,

I recently asked myself the same Question and found an Article where someone made a Test Setup.

You can read through:

http://http://www.qlikfix.com/2011/06/21/testing-the-performance-implications-of-variables-and-label...

Greetz Lukas

[Anonymous]

I can't access the link..?

[luke1986 · Accepted solution]

http://www.qlikfix.com/2011/06/21/testing-the-performance-implications-of-variables-and-label-refere...

Sorry, here you go.

[Anonymous]

Thanks..  I will do the variables.

I have already converted a lot of the text labels.  And i have found a funny observation.  When i change them, the file size of the Qlik application decreases.  so I assume it's storing each label individually, when they are text.  so when i'm inserting the variable. I then save some diskspace, by not having text.  however it's only 10 kb. for 4 tables or so.