Home · Accelerators · Synapse Migration
Migration accelerator
Synapse → Fabric Migration Kit
The complete approach: asset inventory, per-workload remap matrix, parallel-run lakehouse for validation, ADF-to-Fabric pipeline conversion patterns, cutover playbook with documented rollback. Move from Synapse to Fabric without lift-and-shift drag.
Most Synapse-to-Fabric migrations fail one of two ways: they lift-and-shift technical debt into the new platform, or they stall in a never-ending parallel run. This kit prevents both. The remap matrix tells you exactly which Synapse workload maps to which Fabric item. The parallel-run framework gives you a defensible "we're confident" before cutover.
Who this is for
- Data platform teams running Azure Synapse who've been told to move to Fabric.
- Architects who want a defensible migration plan to present to leadership instead of "we'll figure it out."
- Consultancies who need a proven migration framework to deliver against on client engagements.
- Anyone who's tried to plan a Synapse migration and discovered there's no single source of truth on which workload maps where.
You need an existing Synapse workspace and (ideally) some Fabric capacity already provisioned. You don't need to know how Fabric works in depth — the kit teaches you as you migrate.
What's in the box
| File | Type | What it does |
|---|---|---|
discovery/inventory_synapse.ps1 | PS | Pulls every dedicated SQL table, serverless view, Spark notebook, and pipeline from your Synapse workspace into an Excel inventory. |
discovery/remap_matrix.xlsx | XLSX | The canonical Synapse → Fabric mapping. Every Synapse workload type with its Fabric target, effort estimate, and watch-out notes. |
discovery/asset_assessment.xlsx | XLSX | Tactical assessment workbook: which assets to migrate, deprecate, rebuild, or freeze. |
parallel-run/lakehouse_template.json | JSON | Provisions the parallel-run Lakehouse + shortcut configurations to your Synapse data. |
parallel-run/validation_harness.ipynb | Python | Side-by-side query runner: same query on Synapse and Fabric, automated diff, accepted-tolerance report. |
conversion/adf_to_fabric.py | Python | Tool that reads ADF JSON and produces a Fabric pipeline equivalent. Handles 80% of cases automatically; flags edge cases. |
conversion/sql_pool_to_warehouse.sql | T-SQL | DDL conversion patterns: dedicated SQL pool features → Fabric Warehouse equivalents (or workarounds). |
conversion/sproc_to_notebook.py | Python | Stored procedure → parameterized Spark notebook template. Preserves logic, modernizes the runtime. |
conversion/spark_pool_to_fabric.md | MD | Spark pool migration guide: pool configs, library handling, runtime compatibility notes. |
cutover/cutover_playbook.md | MD | Step-by-step cutover plan: communications, freeze windows, switch-over checklist, smoke tests. |
cutover/rollback_procedure.md | MD | When and how to roll back. The plan you hope you don't need but want documented. |
docs/EXECUTIVE_BRIEFING.pptx | PPTX | Editable 14-slide deck for leadership: why, when, cost, risk, timeline. |
docs/COST_COMPARISON.xlsx | XLSX | Synapse vs Fabric cost model: input your usage, see TCO delta. |
The remap matrix at a glance
The included Excel matrix covers every Synapse asset type. A sample:
| Synapse | Fabric equivalent | Effort |
|---|---|---|
| Dedicated SQL pool (DW100c–DW3000c) | Warehouse | Medium — schema conversion + perf tuning |
| Serverless SQL pool | Lakehouse SQL endpoint or Warehouse | Low — direct mapping for most views |
| Apache Spark pool | Fabric Spark (Lakehouse) | Low–Medium — env + library porting |
| Synapse Pipelines | Data Factory pipelines | Low — JSON-compatible, minor activity renames |
| Lake Database | Lakehouse with schemas | Low — Delta-compatible |
| SQL DW external tables | OneLake Shortcuts | Very low — direct replacement |
| Synapse Link for SQL/Cosmos | Fabric Mirroring | Low — Microsoft-managed switch |
| Power BI inside Synapse | Power BI in Fabric workspace | None — same product, new container |
Pricing
Frequently asked questions
Will the ADF converter handle 100% of my pipelines?
~80%. The remaining 20% are edge cases — custom Web activities, complex parameter passing, deprecated linked services — which the tool flags for manual review. Saves you from the 80% of manual work.
What about my dedicated SQL pool stored procedures?
The kit ships both paths: convert to a Fabric Warehouse stored procedure (closer to lift-shift) or convert to a parameterized Spark notebook (more modern). The remap matrix recommends per case.
How long does parallel run take?
Typical: 4–8 weeks. The validation harness automates the row-count and value-match checks, so engineering time is mostly waiting for cycles to complete, not eyeball-comparing.
What if a workload can't fully migrate?
The matrix is honest. Some Synapse features (e.g., Polybase external tables to specific blob locations) don't have direct Fabric equivalents. The matrix tells you which workloads to redesign vs migrate.
Can I use this on multiple client migrations?
Yes — Team license covers unlimited client engagements. Site license adds a kickoff workshop with us.
Refund policy?
Full refund within 14 days.
The migration plan you wish leadership had handed you
Buy it, run the inventory script today, present the executive briefing tomorrow.