Forum Discussion
Guidance Needed on Push Dataset API Retirement
Hi All,
Previously, we implemented a solution using Push Dataset REST APIs in Power BI Service. From our custom web application backend, we invoke the POST (Add Rows) API to update the dataset, and the visuals are built on top of this dataset as per our requirements.
Example use case:
Users interact with our web application to perform specific workflows and process files. During this process, we extract file metadata such as created date, file name, aging, modified date, etc.
This metadata is then sent via the Post Rows in Group REST API to update the dataset, enabling the reports to reflect the latest information.
Recently, we started seeing a notification in the service indicating: “This feature will be retired soon.”
We also reviewed the official updates from Microsoft Power BI Updates Blogs, where it is mentioned that real-time streaming datasets are being retired.
We have a few questions regarding this:
- Is it still possible to continue using the Push Dataset REST APIs in Power BI Service for our existing solution, considering we have ongoing business requirements?
- The blog mentions the possibility of requesting an exception via a support ticket—has anyone tried this, and does it allow continued use of push datasets in production scenarios?
- The suggested alternatives (such as using Fabric-based real-time solutions) seem to require Microsoft Fabric capacity and a different architecture. This would involve significant redesign of our current implementation.
- Is there any recommended migration path that minimizes rework?
- Are there interim solutions for customers who rely heavily on push datasets?
Any guidance or insights from the community or Microsoft team would be greatly appreciated.
Thanks in advance!
Hi Nagaraj_D,
Thanks for reaching fabric community, will happy to assist.
First The Actual Retirement Timeline
Creation of new streaming models (including push semantic models) remains enabled until October 31, 2027. After that date, creation of new real-time semantic models will no longer be supported. Your existing streaming semantic models will be unaffected. Microsoft Learn
So your current production solution is not breaking imminently you have time to plan.
Q1- Can you continue using Push Dataset REST APIs?
Yes, for now. Your existing push dataset and "PostRowsInGroup" calls will keep working until October 2027. However, creating brand new push datasets will eventually be blocked after that date. Plan migration before then, don't wait until the last minute.
Q2 - Exception via support ticket?
Honestly, based on community experience, exceptions are rarely granted for broad platform retirements like this. Microsoft's official stance is to migrate to Fabric Real-Time Intelligence. I wouldn't rely on an exception as a long-term strategy use the time until 2027 to plan a proper migration instead.
Q3 - Migration path that minimises rework
Here's the honest picture: the retirement of real-time streaming means Fabric Real-Time Intelligence using Eventstream and Eventhouse will replace it, and this requires upgrading to a Fabric capacity. Microsoft Fabric Community There is no way around this if you want the full replacement.
However, for your specific use case (pushing file metadata from a web app backend), the migration is actually simpler than a full streaming scenario:
Recommended Migration Path Minimum Rework:
Your Current Approach Recommended ReplacementPOST rows via Push Dataset REST API POST data to Fabric Lakehouse via REST API or Eventstream custom endpoint Power BI visuals on push dataset Power BI report on Direct Lake Semantic Model over Lakehouse Step-by-step:
- From your web app backend, instead of calling "PostRowsInGroup", write metadata to a Fabric Lakehouse table via the Fabric REST API or a simple Azure Function → Eventstream custom endpoint
- Build your Power BI reports on a Direct Lake Semantic Model on top of that Lakehouse table
- Reports refresh near-real-time without scheduled refresh cycles
This approach keeps your backend API pattern similar, avoids a full streaming architecture overhaul, and works well for file metadata (not high-frequency IoT data).
One Thing to Be Aware Of
Fabric Real-Time Intelligence is specifically built for time-based streaming events with features like auto-indexing and partitioning based on ingestion time. Microsoft Community Hub For your use case (file metadata, not millisecond-level streaming), the Lakehouse + Direct Lake path is likely simpler and cheaper than a full Eventhouse/Eventstream setup.
Summary
Question AnswerCan you keep using Push APIs? Yes, until Oct 2027 Exception via support ticket? Unlikely — don't depend on it Minimum rework migration? Lakehouse REST API + Direct Lake Semantic Model Fabric capacity required? Yes, for any Fabric-based replacement Start the migration planning now don't wait for 2027. Hope this helps!
Hi Nagaraj_D , Thank you for reaching out to the Microsoft Community Forum.
You’re not at immediate risk. Microsoft has deprecated real time streaming (which includes push datasets), but your existing push dataset and PostRowsInGroup calls will continue to work for now. The key point is that this pattern is no longer strategic and is expected to be retired.
On exceptions, I suggest you move to newer architectures rather than extend legacy ones. In practice, support tickets don’t give you a reliable long term path here, they may help short term, but they won’t change the direction of the platform.
There isn’t a true like for like replacement for push datasets. The shift is architectural, instead of pushing rows directly into a Power BI semantic model, I suggest you to land data in a storage layer first (for example, a Fabric Lakehouse or another database) and then build your semantic model on top of that. For your use case (file metadata, not high frequency streaming), you can keep your backend pattern but redirect it to write into a table and point your report to that table (Direct Lake in Fabric or DirectQuery/Import as an interim option).
If you want to minimize rework, a phased approach works best, continue using push datasets in the short term, introduce a proper storage layer in parallel and gradually move reports over. That way you avoid a big bang redesign while aligning with where Microsoft is heading.
Real-time streaming in Power BI - Power BI | Microsoft Learn
Push Datasets - REST API (Power BI Power BI REST APIs) | Microsoft Learn
Power BI output from Azure Stream Analytics - Azure Stream Analytics | Microsoft Learn
Announcing the retirement of real-time streaming in Power BI - Power Community
5 Replies
Hi Nagaraj_D,
Thanks for reaching fabric community, will happy to assist.
First The Actual Retirement Timeline
Creation of new streaming models (including push semantic models) remains enabled until October 31, 2027. After that date, creation of new real-time semantic models will no longer be supported. Your existing streaming semantic models will be unaffected. Microsoft Learn
So your current production solution is not breaking imminently you have time to plan.
Q1- Can you continue using Push Dataset REST APIs?
Yes, for now. Your existing push dataset and "PostRowsInGroup" calls will keep working until October 2027. However, creating brand new push datasets will eventually be blocked after that date. Plan migration before then, don't wait until the last minute.
Q2 - Exception via support ticket?
Honestly, based on community experience, exceptions are rarely granted for broad platform retirements like this. Microsoft's official stance is to migrate to Fabric Real-Time Intelligence. I wouldn't rely on an exception as a long-term strategy use the time until 2027 to plan a proper migration instead.
Q3 - Migration path that minimises rework
Here's the honest picture: the retirement of real-time streaming means Fabric Real-Time Intelligence using Eventstream and Eventhouse will replace it, and this requires upgrading to a Fabric capacity. Microsoft Fabric Community There is no way around this if you want the full replacement.
However, for your specific use case (pushing file metadata from a web app backend), the migration is actually simpler than a full streaming scenario:
Recommended Migration Path Minimum Rework:
Your Current Approach Recommended ReplacementPOST rows via Push Dataset REST API POST data to Fabric Lakehouse via REST API or Eventstream custom endpoint Power BI visuals on push dataset Power BI report on Direct Lake Semantic Model over Lakehouse Step-by-step:
- From your web app backend, instead of calling "PostRowsInGroup", write metadata to a Fabric Lakehouse table via the Fabric REST API or a simple Azure Function → Eventstream custom endpoint
- Build your Power BI reports on a Direct Lake Semantic Model on top of that Lakehouse table
- Reports refresh near-real-time without scheduled refresh cycles
This approach keeps your backend API pattern similar, avoids a full streaming architecture overhaul, and works well for file metadata (not high-frequency IoT data).
One Thing to Be Aware Of
Fabric Real-Time Intelligence is specifically built for time-based streaming events with features like auto-indexing and partitioning based on ingestion time. Microsoft Community Hub For your use case (file metadata, not millisecond-level streaming), the Lakehouse + Direct Lake path is likely simpler and cheaper than a full Eventhouse/Eventstream setup.
Summary
Question AnswerCan you keep using Push APIs? Yes, until Oct 2027 Exception via support ticket? Unlikely — don't depend on it Minimum rework migration? Lakehouse REST API + Direct Lake Semantic Model Fabric capacity required? Yes, for any Fabric-based replacement Start the migration planning now don't wait for 2027. Hope this helps!
Hi Nagaraj_D , Thank you for reaching out to the Microsoft Community Forum.
You’re not at immediate risk. Microsoft has deprecated real time streaming (which includes push datasets), but your existing push dataset and PostRowsInGroup calls will continue to work for now. The key point is that this pattern is no longer strategic and is expected to be retired.
On exceptions, I suggest you move to newer architectures rather than extend legacy ones. In practice, support tickets don’t give you a reliable long term path here, they may help short term, but they won’t change the direction of the platform.
There isn’t a true like for like replacement for push datasets. The shift is architectural, instead of pushing rows directly into a Power BI semantic model, I suggest you to land data in a storage layer first (for example, a Fabric Lakehouse or another database) and then build your semantic model on top of that. For your use case (file metadata, not high frequency streaming), you can keep your backend pattern but redirect it to write into a table and point your report to that table (Direct Lake in Fabric or DirectQuery/Import as an interim option).
If you want to minimize rework, a phased approach works best, continue using push datasets in the short term, introduce a proper storage layer in parallel and gradually move reports over. That way you avoid a big bang redesign while aligning with where Microsoft is heading.
Real-time streaming in Power BI - Power BI | Microsoft Learn
Push Datasets - REST API (Power BI Power BI REST APIs) | Microsoft Learn
Power BI output from Azure Stream Analytics - Azure Stream Analytics | Microsoft Learn
Announcing the retirement of real-time streaming in Power BI - Power Community
Hi Nagaraj_D , hope you are doing great. May we know if your issue is solved or if you are still experiencing difficulties. Please share the details as it will help the community, especially others with similar issues.
Hi, thanks for checking in.
I’m currently exploring the Fabric items and will share updates here once I have more clarity.
Regards,
Nagaraj Devadiga
Hi Nagaraj_D , Hope you're doing okay! May we know if it worked for you, or are you still experiencing difficulties? Let us know — your feedback can really help others in the same situation.
