SSIS to Microsoft Fabric Data Factory: Migration Guide for Reporting Pipelines
Three routes for each SSIS package: keep, bridge or rebuild.
SSIS to Microsoft Fabric is not a single migration. It is a decision made package by package: leave it where it is, run it inside Fabric as a bridge, or rebuild it as a pipeline, Dataflow or notebook. Most estates need all three.
This guide is for SQL Server data teams whose SSIS packages feed reporting. It covers the options, the preview activity and its limits, a mapping table, and what each route costs in capacity. For the wider programme, see our guide to migrating to Microsoft Fabric.
SSIS to Microsoft Fabric: Where Your Packages Run Today
First, a correction to a common assumption: SSIS is not end of life. Packages keep running on SQL Server, and in Azure Data Factory through an Azure-SSIS integration runtime, which supports packages in the SSIS catalog, the file system, Azure Files or MSDB on a managed instance. Nothing forces a move. The case for moving is consolidation, because reporting pipelines, storage and semantic models are easier to run on one platform. See our data platform consolidation guide for that case.
Check one more thing before you plan. Fabric Data Factory does not have an Azure-SSIS integration runtime, and the Execute SSIS Package activity from Azure Data Factory is not available there. Microsoft lists SSIS integration in Fabric as a future capability in its comparison, apart from the preview activity covered below.
Your Options: Keep, Bridge or Rebuild
| Route | Where it runs | Best for | Trade-off |
|---|---|---|---|
| Keep | On-premises SQL Server or Azure-SSIS IR in Azure Data Factory | Stable packages with on-premises or private connectivity, or custom components | Two platforms to run and monitor |
| Bridge | Invoke SSIS Package activity in a Fabric pipeline (preview) | Stable packages with OneLake, public cloud sources and no third-party components | Preview limits; still SSIS code to maintain |
| Rebuild | Pipelines, Dataflow Gen2, notebooks, Warehouse SQL | Complex or fragile logic, and reporting that is changing anyway | Highest effort, simplest end state |
The decision tree below turns this into questions you can answer per package. Packages with on-premises sources, private endpoints or third-party components are the first filter, because they fall outside the preview activity.
Decision tree for choosing a route per package. Illustrative.
Run SSIS Packages in Fabric: The Invoke SSIS Package Activity (Preview)
The Invoke SSIS Package activity in Data Factory for Microsoft Fabric lets you move existing SSIS workloads into Fabric with minimal changes. It is in preview, so confirm its status before you commit to it.
The steps are short. Upload your .dtsx packages, and any .dtsConfig files, to a lakehouse in OneLake. Add the activity to a pipeline and browse to the package. If the package protection level is DontSaveSensitive, supply connection strings and credentials on the Connection Managers tab, and override package properties on the Property Overrides tab. Then run or schedule the pipeline and watch it in the monitoring hub. Turn on logging to write execution logs to OneLake.
Know the preview limits before you plan around it. Only packages stored in OneLake are supported. The activity cannot reach on-premises sources or destinations, it does not support private-network endpoints, and it does not support custom or third-party components. Microsoft documents tutorials for writing from packages to a Fabric SQL database, to OneLake through ADLS Gen2, and to a Fabric Data Warehouse using COPY INTO.
SSIS to Microsoft Fabric Package Mapping Table
Rebuilding means mapping each SSIS building block to a Fabric item. This table is the starting point, and the right target depends on the logic inside each task.
| SSIS element | Fabric equivalent | Note |
|---|---|---|
| Package (control flow) | Data pipeline | Same orchestration model, authored in a Fabric workspace. |
| Data Flow task, simple load | Copy activity or Copy job | Use Copy job for incremental movement. |
| Data Flow task, transformations | Dataflow Gen2, Spark notebook or Warehouse SQL | Pick by logic: low-code, custom code or set-based SQL. |
| Execute SQL task | Script or Stored procedure activity | Against a warehouse or SQL source. |
| Foreach Loop container | ForEach activity | Parallel batches are available. |
| Precedence constraints | Activity dependency paths | Succeeded, failed, completed and skipped. |
| Execute Package task | Invoke pipeline activity | Can call pipelines across workspaces. |
| Send Mail task | Office 365 Outlook activity | Also a Teams activity for alerts. |
| Script task (C#) | Notebook or Functions activity | Needs a rewrite; no direct equivalent. |
| Package parameters and configurations | Pipeline parameters and Variable Libraries | Variable Libraries hold environment constants. |
| Connection managers | Fabric connections | On-premises sources use the on-premises data gateway. |
| SQL Server Agent job | Pipeline schedule or event trigger | Schedules and file events are built in. |
| SSISDB logging | Monitoring hub | Preview activity can also log to OneLake. |
Two mappings need care. Script tasks have no direct equivalent, so budget a rewrite in a notebook or function. And transformations are where effort concentrates: for reporting, aim to land clean tables in a lakehouse or warehouse once, so semantic models read from one source. Our posts on Fabric pipeline enhancements and Fast Copy in Dataflows Gen2 cover the tools you will use.
Connections, Gateways and Private Networks
Connection managers become Fabric connections. For on-premises data, Fabric uses the on-premises data gateway in place of a self-hosted integration runtime, and for private Azure resources it uses a virtual network data gateway. Plan these early, because they are the usual blocker. If a package needs an on-premises source today and you want it to stay as SSIS, it cannot use the preview activity, so rebuild it with a gateway or keep it where it is.
Error Handling, Logging and CI/CD
Rebuild error handling rather than copy it. SSIS event handlers become failure paths on the pipeline, with a Fail activity where you need an explicit stop, and Outlook or Teams activities for alerts. Logging moves from SSISDB to the monitoring hub, which gives cross-workspace visibility.
For CI/CD, Fabric offers Git integration and built-in deployment pipelines for promoting items from development to test to production, instead of SSIS project deployment. Keep connection details and environment constants in variable libraries so the same pipeline runs in each stage.
- SSIS is not end of life, so migrate package by package for consolidation, not by deadline.
- Fabric has no Azure-SSIS IR; the Invoke SSIS Package activity is the preview route.
- The preview needs OneLake-stored packages and excludes on-premises, private-network and third-party components.
- Use the mapping table to rebuild, and plan gateways before anything else.
- Validate against report outputs in parallel before you retire a job.
What Each Route Costs in Fabric Capacity
The preview activity bills SSIS uptime at 1.35 CU hours per vCore. Uptime starts when the first activity in the workspace runs, continues while any activity is in progress, and the runtime stays alive for a fixed 30-minute time to live afterwards. Each workspace is allocated 4 vCores during preview. Pipeline runs and OneLake storage are charged under their own meters.
As an illustration, and reading the rate as per vCore-hour of uptime, a single 20-minute package run keeps the runtime up for about 50 minutes. That is roughly 4.5 CU hours, or about $0.81 at the US pay-as-you-go list rate of about $0.18 per CU-hour. Run daily, that is about $24 a month. Back-to-back runs share one TTL window, so chaining packages is cheaper than scattering them. Confirm the meter in Microsoft's pricing page before you quote it.
Rebuilt pipelines consume capacity through activity runs and data movement instead, so compare against your real workloads on a trial capacity. Our Azure Data Factory cost optimisation post covers the pay-as-you-go side, and the Fabric capacity estimator helps size the capacity.
SSIS to Microsoft Fabric FAQ
Is SSIS end of life?
No. SSIS is not end of life, so there is no forced deadline. The reasons to move are consolidation, cost and a simpler operating model. That makes this a planned modernisation, and each package can follow its own timeline.
Can I run SSIS packages in Microsoft Fabric?
Yes, in preview. The Invoke SSIS Package activity in a Fabric pipeline runs .dtsx packages stored in OneLake, with optional .dtsConfig files. It is a preview feature, so its behaviour and limits can change before general availability.
Does Fabric have an Azure-SSIS integration runtime?
No. Microsoft's comparison lists Azure-SSIS integration runtimes as a future capability in Fabric, and the SSIS activity from Azure Data Factory is not available there. The Invoke SSIS Package activity is the Fabric route instead.
What does the preview activity not support?
Only packages stored in OneLake are supported. It cannot connect to on-premises sources or destinations, it does not support private-network endpoints, and packages that depend on custom or third-party components are not supported.
Should I lift and shift or rebuild?
Use the preview activity as a bridge for packages that fit its limits and are stable. Rebuild where logic is complex, where reporting is changing anyway, or where the package needs on-premises or private connectivity, because those are the cases the preview does not cover.
How do I move SQL Server Agent schedules?
Replace each job with a pipeline schedule or event trigger, and rebuild job-step dependencies as activity dependencies. Run the old and new schedules in parallel until row counts and report outputs match.
Before you start, work through the Microsoft Fabric migration checklist and the Microsoft Fabric business case guide. If you want help classifying your packages, talk to our Microsoft Fabric migration team, or book an SSIS assessment.