Stop Executing Queries Multiple Times
I create a query in Power BI Desktop against a data source. Power BI Desktop insists on executing it multiple (usually three) times. This makes Desktop slow and for back-end data sources exerts additional load that can also incur additional costs.
Note: I am not talking about the scenario where a query is referred to multiple times in M, rather just a simple query loaded into one table in the data model. I assume the engine is checking data types, etc - but the current approach is very slow/painful. Optimisation is needed!
This is unacceptable, running queries multiple times is degradding our server performance.
MS should fix this MAJOR ISSUE.
How is this not addressed yet. We are facing lot of performance issues querying Hive with billions of rows. The power bi file has typically multiple tables used and each of the table executes 3 queries internally.
How is this still not resolved?
REST API calls are sending almost 10 queries for datasets larger than 10k records before loading. (PBI Server desktop Sept. 2019 version)
A decent article/vid on this behavior can be found here:
I think the extra querying of metadata is the most difficult to resolve. The workaround in the vid of manually specifying data structure in M code isn't realistic for most people.
This is becoming a significant concern for us as well. Power BI seems to be putting an unnecessary load on our source systems.
What type of source are you having issues with? I was seeing the behaviour you describe with Excel files on SharePoint and found that defining them as "From Web" rather than "From Excel" means that they are cached and not reread
No information from Power BI team?
At least please let us know why it happens
Not sure if I have the exact same problem as the OP but working on the data model in complex cases is very painful and it needs to be easier to jump between transformations to debug issues without issuing a fresh recall of data
Absolutely ridiculous that this happens. I try to refresh a dashboard, the query for 2 years takes 30 minutes. Three executions of the query (2 of which are unnecessary) take 1 hour 30. Not sure why this is.
[Deleted User] commented
Hi Microsoft team !
What about this really important issue ?
It is not acceptable from customers not to have any answer or solution for this old issue, which is really problematic for us as several users run queries from PowerBi Desktop to our datasources (like Hive) and this become now a real performance issue for our platform.
We really need a solution from your team. Quickly.
Chris Amore commented
How do you monitor this type of activity? With the Prem-logs? We are struggling with refresh performance, timeouts and failure and I would like to see if this is part of our issue. This seems like a good way to make Microsoft customers pay for more capacity.
Mike L commented
If it does not get fixed, it may be the death of the product here. I will need to use something else.
We have been fighting the same issue & agree it's a bigger problem with large datasets. We commonly see duplicate requests issued to the source (via on prem gateway) by an otherwise simple Power BI Online refresh.
Very much needed for big data. Having a query that hits billions of records run multiple times can add hours in processing time. We currently have to write the data to a temp table so we can bring it into Power BI, but only one dashboard is using that particular dataset, so...