Forum Discussion
Advocate IPassing parameter to dataflow from Power BI desktop
Advocate IThanks all for the responses!
I initially thought of passing parameters from the dataflow to the source and have the fact table splitted based on some category (ex - country) & consume those separate dataflow entities from power bi report based on requirement. That is of course still possible using ODBC connector (datasource - amazon redshift). But while trying the native redshift connector of Power BI (which has a significant edge in terms of performance over ODBC) - I found no option to pass a sql query while importing! Hence thought of confirming if there is any way to pass parameter to the flow itself from the report.
Anyone knows how to pass parameter from a dataflow to redshift using the native redshift connector 😊? Or I should create a separate post for that??
Thanks for your patience!
Regards,
Rishi
Super UserThink of it from the perspective of your report. Your report connects to the partitions of your dataset and to the DQ connections (if you have any). So your data needs to be present in either. The fact that your dataset feeds from a dataflow is not actually relevant. Well, it means that instead of using a single stage ETL you have multiple stages.
No matter what you do in the earlier stages, eventually your data has to reside in the partitions. One way to improve partition performance would be incremental refresh. But like in SSAS you can decide yourself what the partitions should be.
- rishirajdeb
Thanks lbendlin for the response.
As per my understanding the primary use of dataflows are to maintain a single point of truth and reusability of data transformations - by separating the ETL layer from the data models (datasets). And that is the exact business requirement that we have!
Apart from that as I said, we need to be able to manage large volume of data with a pro license (data needs to be imported as direct query not suitable for our use cases). All data we are importing would change & there is no static part - so don't think incremental refresh would be of much use.
Thanks,
Rishi
- lbendlin
"the primary use of dataflows are to maintain a single point of truth and reusability of data transformations"
Meh. That's what your corporate data stewardship system/data warehouse is for (Azure Purview/CDS/Dataverse and its competitors). Dataflows are not adequate for that - for example they have no guaranteed survivability. You will have scenarios where your dataflow borks down and you have to start over. Not something you want to experience with your data warehouse.
"All data we are importing would change & there is no static part"
Yes, that's a tough one. You need differential refresh and/or the equivalent of CDC. The more people need this the more pressure we can exert on Microsoft to finally support it.