Forum Discussion

DouweMeer's avatar
DouweMeer
Icon for Impactful Individual rankImpactful Individual
6 years ago
Solved

Optimize Service memory usage

So I have a report which has an incremental refresh on a Salesforce source. This Desktop report is published to Service which, fully loaded is approximaly 9GB on Service. This report is (currently) r...
Tips and Tricks
  • JirkaZ's avatar
    JirkaZ
    6 years ago

    DouweMeer Exactly. And btw. the book "The definitive guide to DAX" explains the engine behind Power BI really well.

DouweMeer's avatar
DouweMeer
Icon for Impactful Individual rankImpactful Individual
6 years ago

JirkaZ 

That sounds worrying. So if I have the statement:

calculate ( countrows( selectcolumns ( A , "test" , A.A ) ) , userelationship ( A.B , B.A ) )

You're suggesting the whole table created after an inner join of A and B is stored in RAM during the time the user is accessing the dashboard?

JirkaZ's avatar
JirkaZ
Icon for Solution Specialist rankSolution Specialist
6 years ago

DouweMeer Exactly. And btw. the book "The definitive guide to DAX" explains the engine behind Power BI really well.

  • DouweMeer's avatar
    DouweMeer
    Icon for Impactful Individual rankImpactful Individual
    6 years ago

    JirkaZ 

    I thought the process would be that A and B are pulled into memory to generate the selectcolumns statement. After this is done, the selectcolumns would be stored in cache for further calculation instead of whole of A and B. 

    What I noticed in the past is when you do not use selectcolumns as an intermediair, the calculation would take enourmous amount of times if the table you're calculating on has a relationship with another large table. But maybe that's because I didn't used calculate around selectedcolumns...

    Sounds like I have to dive deep in the VertiPaq storage engine :).