Microsoft Fabric Consultant Guide: OneLake, Lakehouse Setup and Migration From Synapse

Microsoft Fabric brings data engineering, data warehousing, real-time analytics and Power BI into one platform. For mid-market businesses already using Power BI or Azure Synapse, it raises an obvious question: should we move, and if so, how?
This guide covers the three topics we discuss most often with clients considering Fabric: OneLake, lakehouse setup and migration from Synapse. It is written for finance, operations and IT leaders who need to make a sound decision, not just for engineers.
1. What Microsoft Fabric is
Fabric is a software-as-a-service analytics platform. Instead of stitching together separate Azure services for storage, pipelines, warehousing and reporting, Fabric provides them in one place with one capacity and one security model.
Core Fabric workloads include:
Data Factory for pipelines and dataflows
Data Engineering with lakehouses and Spark notebooks
Data Warehouse for SQL-based warehousing
Real-Time Intelligence for streaming and event data
Power BI for semantic models and reports
Data Science for machine learning
Copilot and data agents for AI-assisted analysis
All of these share one storage layer: OneLake.
2. OneLake explained
One copy of the data
OneLake is a single, organisation-wide data lake built into Fabric. Every Fabric tenant gets one. Lakehouses, warehouses and other items store their data in OneLake in the open Delta Parquet format.
Why it matters
No duplicate copies. The same table can be read by Spark, SQL and Power BI without exporting it.
Shortcuts. You can point to data in Azure Data Lake Storage, Amazon S3 or other OneLake locations without moving it.
Direct Lake mode. Power BI can read Delta tables in OneLake directly, giving import-like speed without scheduled copy refreshes.
Central governance. Access, sensitivity labels and lineage apply across items.
Organising OneLake
OneLake is organised by workspaces. A practical structure separates layers and domains, for example:
Finance Bronze, Finance Silver, Finance Gold
Sales Bronze, Sales Silver, Sales Gold
Or one workspace per layer with domain folders. The key is to agree the structure early and stick to it.
3. Lakehouse setup
Lakehouse or warehouse?
Lakehouse | Warehouse | |
Main language | Spark (Python, SQL) plus SQL endpoint for reading | T-SQL |
Best for | Mixed data, files, data engineering | Structured data, SQL-first teams |
Write via | Spark, pipelines, dataflows | T-SQL, pipelines |
Many mid-market businesses use both: a lakehouse for landing and transforming data, and a warehouse or lakehouse gold layer for reporting.
The medallion pattern
Bronze: raw data as received from source systems
Silver: cleaned, deduplicated and standardised data
Gold: business-ready facts and dimensions for reporting
Step by step
Plan capacity. Start with an F SKU sized for your workloads. Many mid-market businesses begin at F2 to F16 and scale as needed.
Create workspaces for each layer or domain, with clear roles.
Ingest data using Data Factory pipelines, Dataflows Gen2 or mirroring for supported databases.
Transform with notebooks, SQL or dbt into silver and gold tables.
Build semantic models on gold tables using Direct Lake where possible.
Apply governance: endorsement, sensitivity labels and access controls.
Monitor with the Fabric Capacity Metrics app.
Mirroring
Fabric mirroring replicates supported databases, such as Azure SQL Database, Snowflake and others, into OneLake in near real time. It reduces the need to build and maintain ingestion pipelines for those sources.
4. Migrating from Azure Synapse
Why teams move
One platform and bill instead of several Azure services
Direct Lake for faster Power BI performance
Built-in Copilot and data agents
Microsoft's product investment is now focused on Fabric
What maps to what
Synapse component | Fabric equivalent |
Dedicated SQL pool | Fabric Warehouse |
Serverless SQL pool | Lakehouse SQL analytics endpoint |
Spark pools | Fabric Spark and notebooks |
Synapse pipelines | Data Factory pipelines in Fabric |
ADLS Gen2 storage | OneLake, or shortcuts to existing ADLS |
Migration approach
Inventory. List pipelines, SQL objects, notebooks, data volumes and Power BI dependencies.
Assess compatibility. Some T-SQL features in dedicated pools behave differently in Fabric Warehouse. Microsoft provides migration guidance and tooling to help assess and convert SQL objects.
Use shortcuts first. Point OneLake shortcuts at existing ADLS data so Fabric can read it without moving it.
Migrate in waves. Start with one domain, such as finance, and run it in parallel.
Reconcile. Compare outputs between Synapse and Fabric before switching reports.
Cut over and decommission. Repoint Power BI models, then switch off Synapse resources to stop paying twice.
Risks to manage
Running both platforms longer than needed
Underestimating T-SQL rewrites
Undersizing capacity and seeing throttling
Skipping reconciliation and losing trust in the numbers
5. Is Fabric right for you?
Fabric is a strong fit if you:
Want one platform instead of several services
Are planning AI use cases on your business data
Have Synapse or older Azure workloads to modernise
It may be less urgent if your current platform works well, costs are stable and you have no plans to expand.
6. How GrowthBI helps
GrowthBI is a Melbourne-based Microsoft Fabric and Power BI consultancy for mid-market businesses with $30M to $500M in revenue. We design OneLake structures, build lakehouses and semantic models, and plan Synapse migrations that reconcile before cut-over.
Pricing
Project pricing: $5,000 to $30,000 AUD per project, depending on scope and data sources.
Ongoing support: from $2,000 per month.
Timeline
Most Fabric setup and migration phases take 1 to 6 weeks, depending on how much data preparation and conversion is needed.
Read more about our data modernisation services.
Frequently Asked Questions
What is Microsoft Fabric?
Fabric is Microsoft's software-as-a-service analytics platform. It combines data integration, engineering, warehousing, real-time analytics, data science and Power BI with a single storage layer called OneLake.
What is OneLake?
OneLake is a single, organisation-wide data lake built into Fabric. All Fabric items store data there in Delta Parquet format, so different tools can use the same data without copies.
Should we use a lakehouse or a warehouse in Fabric?
Use a lakehouse for data engineering, files and Spark-based work. Use a warehouse for SQL-first teams and structured data. Many businesses use both in a medallion design.
Can we migrate from Azure Synapse to Fabric?
Yes. Synapse dedicated pools map to Fabric Warehouse, Spark pools to Fabric Spark and pipelines to Fabric Data Factory. A phased migration with reconciliation reduces risk.
What Fabric capacity do we need?
It depends on data volume, refresh frequency and number of users. Many mid-market businesses start between F2 and F16 and scale up based on capacity metrics.
Plan Your Fabric Move With Confidence
Fabric can simplify your data platform and speed up Power BI, but only with the right structure from day one. A clear plan for OneLake, lakehouse design and migration avoids costly rework.
Book a free consultation with GrowthBI to discuss Microsoft Fabric.