Hello there,
I am reaching out to request an official clarification regarding the supportability of Azure Data Factory (ADF) Change Data Capture (CDC – Preview) when the source system is an on-premises SQL Server instance.
At the moment, I am encountering conflicting behaviors in the platform as well as inconsistent information across Microsoft Q&A threads, which makes it difficult to determine whether this scenario is officially supported.
My Current Environment and Symptoms
On-premises SQL Server 2022
A fully installed and configured Self-Hosted Integration Runtime (SHIR)
ADF Linked Service to on-prem SQL Server through SHIR (connectivity test successful)
ADF Linked Service to Azure SQL (using Managed Identity, also successful)
However, when I attempt to create a Change Data Capture (CDC) resource in ADF:
The Source Linked Service dropdown does not show any Linked Services that use SHIR
Only cloud-based connections (such as Azure SQL) appear, so I cannot select my on-prem SQL Server as a source for CDC
Some Microsoft engineers (2024–2025 responses) indicate that
ADF CDC (Preview) does support on-premises SQL Server through SHIR,
provided that:
The Data Factory is in Live mode, and
The SHIR-based Linked Service is pre-created
They describe the missing dropdown option as a __UI limitation or bug__, not a functional restriction.
__Other official responses__ state that:
__ADF Mapping Data Flows do not support SHIR__,
implying potential CDC limitations or architectural constraints.
__In Microsoft Fabric Data Factory__, Microsoft explicitly states:
__Fabric CDC does not support on-premises data sources at all__
Since I am using classic Azure Data Factory (not Fabric), the above information is contradictory, and it is unclear which rules apply to my environment.
Questions Requiring Official Confirmation
I kindly request definitive answers from the Azure engineering/support team for the following:
1. Does Azure Data Factory (classic, non-Fabric) officially support using CDC (Preview) with an on-premises SQL Server as the source?
If yes, please provide official documentation or prerequisites
(SQL Server version requirements, IR type, network requirements, whether Private Endpoint is required, etc.)
If no, please confirm that this scenario is unsupported and explain the design constraints.
2. Can ADF CDC (Preview) be used with a Self-Hosted Integration Runtime (SHIR)?
If yes, please advise why SHIR-based Linked Services do not appear in the CDC creation UI,
and how to resolve this issue.
If no, please clarify whether this is a known limitation, a bug, or an intentional design.
3. If CDC cannot be used with on-prem SQL Server, what is the recommended Microsoft architecture?
For example:
Azure IR + Private Endpoint to on-prem SQL?
Using Copy Activity with watermark/CDC functions instead of the CDC preview feature?
Staging data in cloud storage before CDC/Dataflow processing?
Or is CDC fundamentally not designed for on-prem sources?
Why This Clarification Is Critical
Due to the inconsistency in available information, I am unable to determine whether:
My issue is caused by a configuration/UI problem
or
The scenario (on-prem SQL + ADF CDC) is fundamentally unsupported
Your guidance will determine whether we should continue with CDC or redesign our data ingestion architecture using incremental copy patterns.
Thank you very much for your assistance. I look forward to your clarification.
Best regards,
Dear Azure Support Team,
I am reaching out to request an official clarification regarding the supportability of Azure Data Factory (ADF) Change Data Capture (CDC – Preview) when the source system is an on-premises SQL Server instance.
At the moment, I am encountering conflicting behaviors in the platform as well as inconsistent information across Microsoft Q&A threads, which makes it difficult to determine whether this scenario is officially supported.
My Current Environment and Symptoms
On-premises SQL Server 2022
A fully installed and configured Self-Hosted Integration Runtime (SHIR)
ADF Linked Service to on-prem SQL Server through SHIR (connectivity test successful)
ADF Linked Service to Azure SQL (using Managed Identity, also successful)
However, when I attempt to create a Change Data Capture (CDC) resource in ADF:
The Source Linked Service dropdown does not show any Linked Services that use SHIR
Only cloud-based connections (such as Azure SQL) appear, so I cannot select my on-prem SQL Server as a source for CDC
Some Microsoft engineers (2024–2025 responses) indicate that
ADF CDC (Preview) does support on-premises SQL Server through SHIR,
provided that:
The Data Factory is in Live mode, and
The SHIR-based Linked Service is pre-created
They describe the missing dropdown option as a __UI limitation or bug__, not a functional restriction.
__Other official responses__ state that:
__ADF Mapping Data Flows do not support SHIR__,
implying potential CDC limitations or architectural constraints.
__In Microsoft Fabric Data Factory__, Microsoft explicitly states:
__Fabric CDC does not support on-premises data sources at all__
Since I am using classic Azure Data Factory (not Fabric), the above information is contradictory, and it is unclear which rules apply to my environment.
Questions Requiring Official Confirmation
I kindly request definitive answers from the Azure engineering/support team for the following:
1. Does Azure Data Factory (classic, non-Fabric) officially support using CDC (Preview) with an on-premises SQL Server as the source?
If yes, please provide official documentation or prerequisites
(SQL Server version requirements, IR type, network requirements, whether Private Endpoint is required, etc.)
If no, please confirm that this scenario is unsupported and explain the design constraints.
2. Can ADF CDC (Preview) be used with a Self-Hosted Integration Runtime (SHIR)?
If yes, please advise why SHIR-based Linked Services do not appear in the CDC creation UI,
and how to resolve this issue.
If no, please clarify whether this is a known limitation, a bug, or an intentional design.
3. If CDC cannot be used with on-prem SQL Server, what is the recommended Microsoft architecture?
For example:
Azure IR + Private Endpoint to on-prem SQL?
Using Copy Activity with watermark/CDC functions instead of the CDC preview feature?
Staging data in cloud storage before CDC/Dataflow processing?
Or is CDC fundamentally not designed for on-prem sources?
Why This Clarification Is Critical
Due to the inconsistency in available information, I am unable to determine whether:
My issue is caused by a configuration/UI problem
or
The scenario (on-prem SQL + ADF CDC) is fundamentally unsupported
Your guidance will determine whether we should continue with CDC or redesign our data ingestion architecture using incremental copy patterns.
Thank you very much for your assistance. I look forward to your clarification.
Best regards,
Lex