Analysis Services - Connect live vs. import help

[powerbi43]

If I connect to my Analysis services database "live", Power BI desktop works just fine.  My tabular model/database is quite small and is only holding about 300 MB of data in-memory.

However, if I choose to "import" that same data into Power BI desktop, it fails with memory errors.  It will stick on a row import of 4096 and the RAM usage on the server will steadily tick up until all the RAM (32 GB) is taken up and I receive a memory error within Power BI desktop.

How could "import"(ing) the data into Power BI desktop possibly be 100 times larger than the data that is held in-memory?  

[Kramnn01]

Hello!

Hello! I also have the same problem. When selecting more than one table from the tabular model, Power BI cross joins the tables, increasing memory a lot. I read in another topic that it is a failure of Power BI and in this case, we must load table by table until the entire model is imported. 

[des]

 Hi what topic did you read this in? could you post a link please?

thanks,

Des

[des]

I can confirm this is what was happening to me, why It cross joins everything seems non-sensicle to me but it still does that even now.

[des]

Hi   @powerbi43  I have experienced this exact same issue, did you ever get this resolved?  If so what was the issue?

[powerbi43]

No, I never resolved the issue

[v-huizhn-msft]

Hi @powerbi43,

When connecting live, no data from the Tabular model is imported into Power BI Desktop. Each time you interact with a visualization, Power BI Desktop queries the Tabular model and calculates the results you see. So it runs quickly and just the pbix file is holding about 300 MB.

When you connect with import model, all the items you selected like tables, columns, and measures in your Tabular model or perspective will be loaded into a Power BI Desktop model. So RAM usage on the server will steadily tick up. You can use Power BI Desktop’s advanced Query Editor to further filter what you want. 

Please review this article:Analysis Services Tabular data in Power BI Desktop.

Best Regards,
Angelia

[powerbi43]

Thank you for the reply.  I had read that article.  Something is definitely amiss in my situation.  The entire set of in-memory data in SSAS is only 300 MB.  How could importing that same set of data into my local Power BI desktop make my server RAM usage jump from 3 GB to almost 32 GB and throw an error?  Keep in mind I hadn't createed any visualations in Power BI desktop.  I was just trying to "load" the data after doing my initial connection to SSAS. 

I thought I had perhaps defined my relationships incorrectly in my tabular model, but they're fine.  For now my solution is just to connect Live (not sure why you'd import in-memory data anyway).  The default is import, which I used when creating my initial connection.  

[v-huizhn-msft]

Hi @powerbi43,

Please read the knowledge and the article: SSAS TABULAR 2016: DIRECT QUERY VS. IN-MEMORY. 

Thanks,
Angelia