Skip to main content
Announcements
Accelerate Your Success: Fuel your data and AI journey with the right services, delivered by our experts. Learn More
cancel
Showing results for 
Search instead for 
Did you mean: 
Fabiano_Martino_Intelco
Partner - Creator II
Partner - Creator II

Prevent users from combining specific dimensions

Hi,

we have users with the full level authorization, so they can create their own charts.

In our data model we have two fact tables (say purchases and sales) and sometimes the users will create charts picking dimensions and measures from both tables.

This creates a spike in the server RAM, because the specific combination of dimensions and measures creates a cross join.

Can we use a preemptive control to avoid this situation? It would be nice to put a memory limit on the specific charts or, even better, to prohibit specific dimensions and measures combinations.

Thanks

Fabiano

Labels (2)
1 Solution

Accepted Solutions
marcus_sommer

I suggest not to cure the effects else the cause - which means to adjust the data-model. Officially recommended is a star-scheme with a single fact-table and n surrounding dimension-tables.

View solution in original post

3 Replies
rwunderlich
Partner Ambassador/MVP
Partner Ambassador/MVP

You can detect these multi-table combinations using QSDA Pro after the fact. But I don't have an idea as to how you can prevent them from getting created in the first place. 

-Rob

Fabiano_Martino_Intelco
Partner - Creator II
Partner - Creator II
Author

Hi @rwunderlich 

QSDA Pro is indeed a useful tool.

In my specific scenario I would like to have a check before the situation arises.

The function GetObjectDimension() seems interesting, but it requires the index of the specific dimension to be tested.

Since I won't know beforehand how many dimensions the user will select I have a problem here.

It would be nice to have a function returning the complete list of dimension for a table, maybe with a custom separator.

Thanks

marcus_sommer

I suggest not to cure the effects else the cause - which means to adjust the data-model. Officially recommended is a star-scheme with a single fact-table and n surrounding dimension-tables.