Power BI - Snowflake Direct Query Too Slow

[JOSEPIVIDAL]

Hi,

I am using the Direct Query Mode with the SnowFlake connector to query a 650M rows & 50 Columns database.

I am getteing serious performance issues to load visuals on Power Bi, with a Matrix of 30 calculated fields taking up to 15 minutes to load. 

Looking into the Snowflake "history" of querys, I see that the problem seems to be that Snowflake is receiving / Power BI Sending the queries one after the order, so when I refresh a visual, SF receives a sequence hundres of queries in blocks of 2 / 3 per second. All of them seem to get calculated in mileseconds, but as they get in blocks of 3 (per second) it takes for ages. I have Large Datawarehouse with 8 clusters, which is supposed to be more than enough, in fact no more than 2 clusters have ever been needed, that makes me think the problem is coming from Power Bi.

These are the settings I have in Power BI:

I also have enabled the Parralel loading of tables option.

Tried to look everywhere for this issue, but still have not turn a good datables into a good analysis tool because of this.

Thanks in advance for your help!

Jose

[abhilashyerasi]

Hi @JOSEPIVIDAL 

Its been quite long on this.

I would like to check if you found something on this. I have raised a ticket to support.

If you have any potential solution would be really benificial for the community.

Thanks in Advance.

[Anonymous]

Hi, I'm having the same issue here. Did you find a solution for this? Thanks.

[Anonymous]

Hi, did you guys have a solution for the problem?. I landed in a similar issue

Thank you

Pavan

[amitchandak]

@JOSEPIVIDAL , Avoid bidirectional joins. Keep only required visual on the page

refer to these Guidelines: https://docs.microsoft.com/en-us/power-bi/guidance/directquery-model-guidance

refer  :https://www.thebiccountant.com/speedperformance-aspects/

[JOSEPIVIDAL]

Thanks for your help @amitchandak but I am trying to find out why is Snowflake getting those queries' start time with that sequence of 2/3 queries per second, that seems to be the reason for making all the process so slow as snowflake itself seems to process the data very quickly.