Data type emulation
Last updated
Last updated
Data type differences are a major source of incompatibility when migrating from Synapse Analytics to Fabric Warehouse.
Synapse supports a broad set of SQL Server-compatible types, including Unicode and legacy data types, while Fabric Warehouse uses a simplified, optimized type system tailored for distributed analytics.
Many commonly used types in Synapse — such as NVARCHAR
, NTEXT
, and DATETIMEOFFSET
— are not supported or behave differently in Fabric. These differences can lead to deployment failures or unexpected runtime behavior if not properly handled.
SQL Tran automatically resolves these incompatibilities by converting unsupported Synapse data types into their closest Fabric equivalents.
For example, NVARCHAR
is translated into VARCHAR
with doubled length to maintain character capacity under UTF-8 encoding, and MONEY
is converted to a fixed-precision DECIMAL
type.
This intelligent translation ensures that migrated schemas retain their original behavior without requiring manual intervention — a process that would be time-consuming and error-prone if done by hand.
By automating data type adjustments across all objects in the database, SQL Tran significantly reduces migration effort and risk, enabling fast and reliable transitions to Fabric.
(For more information, see the following emulation reference page: )
Synapse Analytics:
Fabric Warehouse (generated by SQL Tran):
In the example shown, SQL Tran automatically transforms the data type NVARCHAR(50)
to VARCHAR(100)
.
The length is doubled to maintain character capacity under UTF-8 encoding, ensuring compatibility with Fabric’s storage system.
All other attributes are carefully adjusted to align with Fabric's requirements.