The ultimate Fabric, Power BI, SQL, and AI community-led learning event. Save €200 with code FABCOMM.
Get registeredEnhance your career with this limited time 50% discount on Fabric and Power BI exams. Ends August 31st. Request your voucher.
For troubleshooting purposes, I occasionally need to run profiler against a PBI dataset model.
This used to be pretty fast in PBI (on a gen1 P1 capacity).
I'm not sure when Microsoft started breaking this functionality, possibly in gen2 premium or in the new "F" sku's. But the profiler is impossible to use nowadays. It locks up all the time, returns strange errors, and doesn't show events that we expect (when running queries).
Login to profiler:
powerbi://api.powerbi.com/v1.0/myorg/GR-IT-DEV-MyCoolStuff
Entra with password
Use dataset name on the second tab.
Here are links that seem to indicate that it should still be working:
https://learn.microsoft.com/en-us/power-bi/guidance/monitor-report-performance
Question - is profiler fully supported on F64 sku?
Solved! Go to Solution.
Hi @dbeavon3 ,
Thanks for your detailed response and for sharing your observations. I understand your frustration and the need for clarity on Profiler’s support on F64 SKUs.
Regarding Profiler Support on F64 SKU:
Yes, Profiler is supported on F64 SKUs, as XMLA endpoint functionality is included in the F-series SKUs.
However, we don’t have specific documentation confirming whether Profiler has undergone full regression testing on these SKUs. If there are functional gaps or unexpected behavior, raising a support case is the best way to ensure the Microsoft product team is aware and can investigate further.
Additionally, we would like to inform you that SQL Server Profiler is deprecated and will be removed in a future version of SQL Server. While Analysis Services workloads (including Power BI datasets) are still supported, Microsoft recommends using Extended Events (XEvents) instead for more reliable monitoring and troubleshooting.
Given that Profiler is not actively developed, there’s a possibility that changes in the F-series SKU or Fabric backend could impact its functionality over time.
Please refer this links for more details:
SQL Server Profiler - SQL Server Profiler | Microsoft Learn
extended-events-vs-sql-server-profiler
Regarding Profiler Behavior with Semantic Link (Spark Native Connector):
You’re right that queries from Semantic Link should still be executing against the Power BI dataset, likely via DAX or MDX, and those events should be visible in Profiler. There’s no intentional "stealth mode" hiding Spark-generated queries, but it’s possible that queries from this connector are processed differently, affecting how Profiler captures them.
Given Profiler’s deprecated status, it may not handle newer workloads as effectively.
Hope this helps.
If this post helps, then please consider Accept it as the solution to help the other members find it more quickly and a kudos would be appreciated.
Best Regards,
Vinay.
Hi @dbeavon3 ,
Thanks for reaching out in Microsoft Fabric Community and sharing additional details.
A few things could be affecting the results in SQL Server Profiler.
1. First, Cloudflare Warp might be interfering with network traffic, so try disabling it temporarily to see if that makes a difference. Additionally, Profiler works best when monitoring only the most relevant events and tracking too many can slow things down and cause issues. Filtering the events you capture might help.
2. Another key point is ensuring that you're connecting properly. When connecting to a Power BI Premium workspace, make sure to enter the correct database (semantic model) name in Connection Properties > Connect to database. If this isn't specified, you might run into permission errors or missing events. Also, check that XMLA read-only access is enabled on your Premium capacity, as it's required for Profiler to function correctly.
3. It's also important to note that SQL Server Profiler is primarily designed for SQL Server and Analysis Services (SSAS). It captures T-SQL queries for SQL Server and MDX/DAX queries for Analysis Services, including Power BI datasets. However, queries executed via the Semantic Link (Spark native connector for Fabric) may not be fully captured, as Profiler isn't optimized for Spark-based query execution.
4. For troubleshooting Spark-based activity, consider using Fabric's Spark Job Monitoring and the Fabric Monitoring Hub, as they can provide better insights into query performance and execution.
If Profiler still isn’t capturing queries, you might want to try DAX Studio or Performance Analyzer in Power BI, as they can sometimes offer more detailed insights into dataset performance.
Hope this helps. Please let us know for further assistance.
If this post helps, then please consider Accept it as the solution to help the other members find it more quickly and a kudos would be appreciated.
Best Regards,
Vinay.
Hi @v-veshwara-msft
Thanks for taking time to reply.
My primary question is whether profiler fully supported on F64 sku. Do you know if that will be supported when it stops working over time? Do they fully test it, as they are making changes to the back-end (eg. when the "F" sku technology was created, did they do regression testing on their profiler?)
I can open a ticket with Mindtree but it will take several weeks before they send the case to Microsoft. I don't want to go to that effort if the Microsoft PG won't provide meaningful support for their SQL Profiler tool. There are some obvious bugs that I'm seeing and I'm guessing that the support for this tool is poor, or they would have seen the bugs and fixed them on their own initiative for their own reasons.
1. Yes, as I mentioned cloudflare warp is a likely cause of some of the issues in profiler. For example, the events wouldn't come thru in a timely manner, but if I pushed the "pause" button on the profiler, then a bunch of latent events would suddenly appear after-the-fact. Obviously that is not the expected behavior for the tool.
2. Yes, as I mentioned I connect with the dataset name on the second tab.
3. How do you suppose that Spark (native connector) would get data from a PBI dataset WITHOUT sending those events to the profiler? The act of querying a dataset is not altogether unsual, if the remote client is one type of application or another type of application. The queries from Spark are almost certainly normal queries via DAX. Are you proposing that Microsoft introduced some sort of "stealth mode" where they deliberately hide the queries that are issued from Spark in Fabric?
4. I have tried other tools as well. The capabilities of SQL profiler aren't available in the other tools. I'm trying to troubleshoot bugs in Microsoft's native connector for Spark. That is a hard enough job without encountering additional bugs in SQL profiler as well.
Any information would be appreciated, especially about the likelihood of getting support from the Microsoft PG.
Hi @dbeavon3 ,
Thanks for your detailed response and for sharing your observations. I understand your frustration and the need for clarity on Profiler’s support on F64 SKUs.
Regarding Profiler Support on F64 SKU:
Yes, Profiler is supported on F64 SKUs, as XMLA endpoint functionality is included in the F-series SKUs.
However, we don’t have specific documentation confirming whether Profiler has undergone full regression testing on these SKUs. If there are functional gaps or unexpected behavior, raising a support case is the best way to ensure the Microsoft product team is aware and can investigate further.
Additionally, we would like to inform you that SQL Server Profiler is deprecated and will be removed in a future version of SQL Server. While Analysis Services workloads (including Power BI datasets) are still supported, Microsoft recommends using Extended Events (XEvents) instead for more reliable monitoring and troubleshooting.
Given that Profiler is not actively developed, there’s a possibility that changes in the F-series SKU or Fabric backend could impact its functionality over time.
Please refer this links for more details:
SQL Server Profiler - SQL Server Profiler | Microsoft Learn
extended-events-vs-sql-server-profiler
Regarding Profiler Behavior with Semantic Link (Spark Native Connector):
You’re right that queries from Semantic Link should still be executing against the Power BI dataset, likely via DAX or MDX, and those events should be visible in Profiler. There’s no intentional "stealth mode" hiding Spark-generated queries, but it’s possible that queries from this connector are processed differently, affecting how Profiler captures them.
Given Profiler’s deprecated status, it may not handle newer workloads as effectively.
Hope this helps.
If this post helps, then please consider Accept it as the solution to help the other members find it more quickly and a kudos would be appreciated.
Best Regards,
Vinay.
Hi @dbeavon3 ,
Since we haven’t heard from you in a while, we hope the information shared helped clarify your concerns regarding Profiler support on F64 SKUs.
If any of the responses here were helpful, please consider marking the relevant one as the Accepted Solution , this can assist other community members facing similar questions.
Additionally, if you raised a support ticket and received further insights or confirmation from the Microsoft team, we’d appreciate it if you could share those findings here for the benefit of the wider community.
Should you need further assistance in the future, please feel free to create a new post in the Microsoft Fabric Community Forum. We’ll be glad to support you.
We are closing this thread for now.
Thank you for your time and participation.
A couple extra details....
We are using cloudflare "warp", and that is contributing to some of the problem but not all.
I'm trying to monitor activity from the "Semantic Link" connector for spark (aka "spark native connector" for pyspark in fabric). Only a small number of events are coming thru. No queries are seen, but some "Discover" stuff and "Execution Metrics" and "Execute MDX Script". Image.