Forum Discussion

Jack_W_KY's avatar
Jack_W_KY
New Member
1 year ago
Solved

DATE_DIM tables and Direct Query

In my org, we default all cloud data sources to direct query.  My group is sourcing our data in Snowflake sitting on Azure.  We need to use date heirarchy logic in several reports we've built and we ...
Need Help
Tips and Tricks
  • v-sdhruv's avatar
    v-sdhruv
    1 year ago

    Hi Jack_W_KY ,
    You can try these approaches-

    Run Query diagnostics in Power BI desktop

    Go to: Tools > Query Diagnostics > Start Diagnostics

    Check logs for steps that take too long

    If you see long durations between “Data retrieved” and “Data displayed” — it's not Snowflake, it’s post-processing or service-side bottleneck

    Try:
    1.Temporarily using Import Mode for DATE_DIM Only (if Composite Models are enabled)
    2.Move DATE_DIM into Snowflake
    3.Reduce number of visuals, filters, bookmarks
    4.Test different gateway config
    5.Reduce result set by limiting rows/columns

    If the response has addressed your query, please accept it as a solution and give a 'Kudos' so other members can easily find it.
    Thank You

collinq's avatar
collinq
Super User
1 year ago

Hi Jack_W_KY ,

I am not sure that I have seen this problem before.  I am wondering if you have a connection issue with Snowflake that might be slowing things down?  I am also wondering if you have Kerberos security or something that might be slowing it down?

You might want to check the Perfomance Analyzer to see if it helps you find the spot that is an issue.  Or, perhaps use Fiddler to see if it is a network issue.

Jack_W_KY's avatar
Jack_W_KY
New Member
1 year ago

When I look at the performance analyzer I see the DAX and Direct Query are only a few seconds off each other in total run time.  When our IT group pulls the queries sent back to Snowflake, the Direct Query is executing in less than 3 seconds of server time. 

 

Our gateway server is an Onprem solution.  The Admin of that server set the gateway server up so it doesn't cache the data before sending it back to the service in hopes of improving overall processing but I'm still getting poor results in a handfull of reports.