Enter your search text in the box above

Select a result to preview

Peachy Migrations: Sweet Moves from Synapse to Fabric

#fabcon2026 #session #learning #microsoft #fabcon2026-session

Description

Ready to make your analytics sweeter? Learn why migrating to Fabric’s OneLake architecture and AI-driven features beats Synapse for speed, governance, and insights. In this demo-driven session based on real-world migrations, learn practical steps, compatibility fixes, and how tools like the Fabric Migration Assistant make moving to Fabric Data Warehouse as smooth as a Georgia peach.

Key Takeaways

My Notes

Action Items

Slides

Peachy Migrations:
Sweet Moves from
Synapse to Fabric
Rob Carrol
Principal Consultant, Coeo
rob.carrol@coeo.com
https://www.linkedin.com/in/rob-carrol-sql/
Why Migrate?
Migration tools
Lessons learned
Microsoft Fabric
The unified data platform for AI transformation
Data Factory
Analytics
Databases
Real-Time
Intelligence
IQ
Power BI
Fabric Platform
AI
OneLake
Security & Governance
Microsoft Fabric
The unified data platform for AI transformation
Data
Factory
Data
Engineering
Data
Warehouse
Why migrate to Fabric Data Warehouse?
Single source of truth, open storage format
on OneLake
Interactive BI reporting with Power BI and
Direct Lake
Agentic AI as a first-class capability,
enabling business users to chat with data
T-SQL
Intelligent SQL Engine
AI
Serverless compute
Secure and govern the most complex data
on a unified SaaS platform with fewer
moving parts.
Capacity-based cost model and quicker
scale-up/down
Open storage format
in customer-owned OneLake
Rich T-SQL surface area
Why migrate: Data Warehouse Performance
1 TB TPCH - 22 Queries - Warm Cache
1.9x
Runtime
Cost
1.7x
Synapse Gen2
Fabric DW
Synapse Gen2
Fabric DW
Source: Based on benchmarking study conducted by Microsoft Corporation August 2025 using TPC- H like metrics on 1 TB dataset. Runtime and Cost per query were calculated
for Azure Synapse Dedicated Pools DWU1000 and Fabric Data Warehouse F64.
Why migrate: Fabric Data Engineering vs Synapse
Cost
efficiency
SaaS-based Spark offering
Superior
performance
TPC-DC 1TB
Built-in starter pool requiring minimal
Spark expertise
TPC-DC 1TB
Enhanced Spark runtime with Microsoft innovations
(Native engine, high concurrency, etc.)
1.4x
2x
vs Synapse
3.5x
vs Synapse
+
Execution Time
Execution Cost $
vs Synapse
2.4x
vs Synapse
Cross-workspace Lakehouse on OneLake using
open Delta Lake format
Integrated with Fabric data agents for natural
language query and the
OneLake Catalog for governance and
discoverability
+
Native Execution
Engine
Native Execution
Engine
*Synapse Spark in non-US regions typically uses the same VM SKU
as Fabric so there would be less perf diff compared to Fabric w/o NEE
*Synapse Spark in non-US regions typically uses the same VM SKU
as Fabric so there would be less perf diff compared to Fabric w/o NEE
Copilot-assisted capabilities for autonomous
optimization and diagnostics
Source: Based on benchmarking study conducted by Microsoft Corporation 2024 using TPC-DC metrics on 1 TB dataset. Cost Efficiency and Performance were calculated for Azure Synapse Spark and Fabric Data Engineering for comparable SKUs
Why Migrate: Fabric Data Factory performance boost
16,695
Run duration (Seconds)
Dataflow Gen 1
12,194
Dataflow Gen 2 (No perf enhancements)
Dataflow Gen 2 with Modern Query Evaluator
Dataflow Gen 2 with Modern Query Evaluator
and Parallelized Execution
6,415
4,762
2,944
1,601
Benchmark 1
1,753
Benchmark 2
Benchmark #1: NYCTaxi-2017, 1 partition, 9.8GB, 113M rows, 2 added columns
Benchmark #2: NYCTaxi-20xx, 5 partitions, 32.2GB, 365M rows, 2 added columns
Migration Tools
Data Warehouse
Data Engineering
Data Factory
DW Migration Assistant (GA)
•Live connectivity (coming soon)
Discovery and Assessment Tool
Migration Assistant (Public Preview)
NEW
Semi-auto Migration Script
PowerShell Upgrade Script
Migration Assessment Assistant
Fabric Data Factory Migration Assistant
(Microsoft CAT fabric-toolbox)
Cloud Accelerate Factory
3rd Party tools
Data Warehouse Migration Assistant
Data warehouse migration
Data sources
Microsoft Fabric
Built into Fabric
Intelligent assessment
Synapse Dedicated
SQL Pools
Data Warehouse
Code conversion
Data migration
SQL Server
Productivity
OneLake
AI-powered
Performance
Decision-making
Innovation
Migrating to Fabric Data Warehouse
Migration Assistant
Source Database
Synapse Dedicated
SQL Pools
Migration Solution
SQL Databases
DDL Conversion
Microsoft Fabric
Script Deployment
Data Warehouse
ADLS Gen2
Parquet
CSV
Identify a source
and create a
DACPAC file
Migrate the data
warehouse using
the DACPAC file
Review the successfully
migrated objects in the
Fabric data warehouse
Copy the source data
to ADLS Gen2 into
Parquet or CSV format
Load the data to the Fabric data warehouse
using the COPY INTO T-SQL command or your
choice of tool (ADF, Pipeline, Copy Job etc)
DEMO
Data Warehouse
Migration Assistant
Warehouse Migration Assistant Considerations
AI features require Copilot to be activated
• Requires a paid capacity (F2 or higher) - Copilot is not supported in trial capacities
Copy metadata and data
• DACPAC is (currently) required to copy metadata
• Copy Jobs are used to copy the data. For large data copies:
• Check your VNET Data Gateway (if using one) for bottlenecks
• Use CETAS to export the data to ADLS Gen2 then COPY INTO to import data into Fabric
• Efficiently Moving Data from Dedicated SQL Pool to Fabric Data Warehouse | by Mark Pryce-Maher | Medium
Efficiently Moving Data from Dedicated SQL Pool
to Fabric Data Warehouse | by Mark Pryce-Maher | Medium
Warehouse collation
• Migration Assistant creates the new warehouse with a case-insensitive collation
• Once a warehouse or SQL analytics endpoint is created, the collation can’t be changed
Review “State” of migrated objects
• “Adjusted” means metadata was changed during the translation
• Datatypes or T-SQL constructs are changed to ones supported in Fabric Data Warehouse
Review failed objects
• Copilot can suggest fixes, but manual input will be required
Other Migration Considerations
aka.ms/fabricroadmap
T-SQL incompatibilities
• IDENTITY column differences
• Datetime isn’t supported
• HASHBYTES should use SHA2 instead of SHA1
ISOLATION LEVEL
• Fabric Warehouse supports SNAPSHOT isolation only
Run an assessment of the source code
• Assess T-SQL code before migration to identify issues and anti-patterns
• Fix as many of these prior to migration as possible
DEMO
Fabric Warehouse
Code Compatibility
Assessment
What about Synapse Serverless SQL pools?
Migration Assistant only supports Dedicated SQL pools
• For Serverless Pools, assess external tables and views by source format (Delta, Parquet, CSV)
Option 1 - Keep ADLS Gen2 as default storage and use shortcuts
• Create OneLake shortcuts to your existing ADLS Gen2 paths
• For Delta, create the shortcut under /Tables
• Shortcuts for CSV/JSON/Parquet created in /Files
• N.B. These /Files shortcuts won’t be automatically discovered by the SQL endpoint
Option 2 - Move data to OneLake as default storage
• Migrating into OneLake reduces external dependencies, simplifies governance
Do you even need a warehouse?
• Start with Lakehouse + SQL endpoint (plus views) to replace Serverless querying.
• Add a Warehouse only when you need writable T-SQL or Warehouse-specific features as the
SQL Analytics endpoint is read-only
Data Engineering migration
Azure Synapse workspace
Spark pools
Libraries
Custom pools
Spark Job
Definitions
Notebooks
LakeDatabases
Hive Metastore
Properties
Environment
Notebooks
SJDs
Lakehouse
Fabric workspace
Data Engineering items migration
Azure Synapse
Compute
Resources
Data
HMS
Fabric Spark
Spark pools
Starter and Custom Spark pools
Spark configurations (pool level)
Environments
Libraries (pool level)
Environments
Notebooks (Python, Scala, Spark SQL, R, .NET)
Notebooks (Python, Scala, Spark SQL, R)
Spark job definitions (Python, Scala, .NET)
Spark job definitions (Python, Scala, R)
Notebook pipeline activity
Fabric pipeline Notebook activity
Spark Job Definition pipeline activity
Fabric pipeline SJD activity
ADLS Gen2 as main storage
• OneLake as main storage
• Use shortcuts to ADLS Gen2
Internal
External (Azure SQL DB based)
• Use shortcuts > Delta tables only
• Lakehouse /Tables > Delta & non-Delta tables
NEW
Data Engineering Migration Assistant (Public Preview)
DE Items Migration
Azure Synapse
Spark & Lakedatabase
•
Notebook
•
Spark Job Definition
•
Spark pools
•
Spark Configuration
•
Lake Database
Microsoft Fabric
Fabric Native Experience
Spark & Lakehouse
Built-in validation rules
Migration monitoring
Detailed migration report
•
Notebook
•
Spark Job Definition
•
Environment
•
Lakehouse
•
OneLake
Retain items level association
DEMO
Data Engineering
Migration Assistant
How to Migrate: Additional Considerations
Spark Code & Runtime
• Replace mssparkutils with notebookutils and validate runtime versions
• Move Linked services and data source connections
• Cloud connections, On-Prem Data Gateway, VNet Data Gateway
• Synapse ADLS Gen2 URLs (abfss) should be updated to read/write through OneLake or shortcuts
• spark.table(“schema.table”), df.write.format(“delta”).saveAsTable(“schema.table”)
Libraries & Dependencies
• Rebuild any custom libraries as Fabric Environments. Test any upgraded libraries for syntax changes
Security & Governance
• Assing workspace RBAC roles and access policies
• (Managed) private endpoints and VNets
• Repoint Power BI, apps and reports to use Fabric endpoints
Validation
• Run workloads in parallel and validate schema/data.
• Run a code review and look for deprecated APIs
Data Factory Migration Considerations
Feature
Fabric vs ADF
Integration runtimes
Fabric uses cloud-based compute by default; ADF requires configured IRs.
Pipelines
Fabric includes SaaS-based activities and uses different JSON definitions.
Linked services
Fabric replaces Linked Services with in-activity Connections.
Datasets
Fabric removes datasets - data properties are defined inline.
Dataflows
Fabric uses Power Query; ADF Data Flows use a different engine & language.
Triggers
Fabric uses Activator framework vs ADF standalone triggers.
Debugging
Fabric removes separate debug mode for simplified authoring.
Change Data Capture
Fabric uses Copy jobs instead of CDC artifacts.
Azure Synapse Link
Mirroring replaces Synapse Link for replication.
SSIS
Fabric doesn’t support SSIS IRs but can invoke ADF for SSIS execution.
Invoke pipeline
Fabric enhances Execute pipeline with cross-platform invocation.
Pipeline migration readiness assessment in Synapse
・ Assess your Synapse
pipelines for “Fabric
readiness”.
・ Expand results to view
details of assessment
Click “Start assessment
(preview)” to run a migration
readiness assessment of your
Synapse pipelines.
・ Export results to plan your
migration.
DEMO
Migration
Assessment Tool
DEMO
Fabric Data
Factory Migration
Assistant
Pipeline migration tips
Run the assessment tool
• Plan for any incompatibilities, and pipelines that need to be manually
migrated. For example, invoke Synapse notebook tasks, sFTP tasks
Make sure Synapse code is in a repo
• Download latest ARM template from your Synapse Repo
(TemplateForWorkspace.json)
Create connections before migration
• Use the same names and the Fabric Data Factory migration assistant will map
them automatically
Start small and test
Automate where possible
Use git integration and CI/CD
• Migrate a few pipelines at a time and test the output
• Use the automated migration tools as much as possible
• AI agents were used to refactor some incompatible code
• Once migrated, enable git integration for your Fabric workspace and follow
CI/CD release best practices for your pipelines in Fabric
Cloud Accelerate Factory
Jumpstart Azure projects with zero-cost deployment from Microsoft experts
Factory benefits
Get zero cost
deployment assistance
Get Microsoft assistance
to deploy 30+ Azure
services via joint delivery
with an Azure partner no
matter the project size
Accelerate with
industry best practices
Leverage proven strategies
developed over thousands
of customer engagements
to complete deployments
quickly
Unlock and scale
your resources
Prioritize your funding
and skilled resources for the
more advanced
components or projects to
boost business impact
Third-party Fabric workloads
Kanerika Inc.'s migration
accelerator helps move your
Azure Data Factory (ADF) and
Synapse pipelines into Fabric
Data Factory (FDF)
• Assess & map ADF/Synapse
pipelines
• Convert activities for Fabric
optimization
• Configure integrations across
lakehouse, warehouse & semantic
models
• Validate, test, and optimize for peak
performance
Third-party Fabric workloads
• Announced this week
• Tool for migrating SQL Server
and Synapse databases to
Microsoft Fabric
• FREE plan supports:
• SQL Server, Azure SQL
Database, Azure SQL Managed
Instance, and Synapse
Analytics Dedicated SQL Pool.
• Powerful testing framework.
• Premium plan adds:
• Automated testing
• Migration from other database
sources
Microsoft Fabric Migration Resources
Sound off.
The mic is all yours.
Influence the product roadmap.
Join the Fabric User Panel
Join the SQL User Panel
Share your feedback directly with our
Fabric product group and researchers.
Influence our SQL roadmap and ensure
it meets your real-life needs
https://aka.ms/JoinFabricUserPanel
https://aka.ms/JoinSQLUserPanel
How was
the session?
Complete Session Surveys in
for your chance to WIN
PRIZES!