Supported data

General Validation compares Tables that already live in Microsoft Fabric. You pair a source Table with a target Table, add Tests, and run them in your own Fabric workspace. This page lists what you can pair, how each kind of Table is read, and the limits that apply to a Table Pair. The twelve Tests are described on Tests, and the steps for pairing Tables are on Create Table Pairs.

Supported inputs

A Table Pair joins one source Table and one target Table. Each side is a Fabric Lakehouse Table or a Fabric Warehouse Table, chosen from a workspace your Microsoft account can access. Creating the pair saves only the Table reference; no rows are copied into General Validation.

Input	How it is read
Fabric Lakehouse Table	Read as a Delta Table. Schema-enabled and schema-less Lakehouses are both supported. A Table reached through a shortcut is read at its logical path inside the selected Lakehouse; General Validation never follows the shortcut to its physical destination, and Fabric controls access to the data behind it. The Table Pair can pin the Table to a Delta version.
Fabric Warehouse Table	Read through SQL with your own SQL permissions, so the user who runs validation needs SQL SELECT on the Table as well as Read on the Warehouse item; SQL permission alone does not grant the OneLake access used to browse columns. A denied SQL read reports the Test as an error; there is no fallback to storage. Warehouse Tables cannot be pinned to a version.

Any combination of the two forms a Table Pair: Lakehouse to Lakehouse, Lakehouse to Warehouse, Warehouse to Lakehouse, or Warehouse to Warehouse. Tables are listed with their schema, so identical Table names in different schemas stay distinct. The dedicated General Validation workspace holds Results only; it is never a source or target and does not appear in the workspace list.

The 12 validation functions

Every Test applies one of twelve fixed functions to a Table Pair: seven aggregate comparisons, two row comparisons, and three set comparisons. Tests lists each function, what it compares, how comparisons behave, and how to add one.

Limits

Limit	Value
Join keys per Table Pair	1 to 16 source-to-target mappings; needed only for `VALUE` and `OUTER_VALUE`
Filters per side	Up to 32 on the source and up to 32 on the target
Values in one In list filter	1 to 100
Active Table Pairs	Set by your plan; see Plans and trial

Filters use the operators Equals, Does not equal, Less than, Less than or equal, Greater than, Greater than or equal, In list, Is null, and Is not null. Source filters and target filters are independent lists, and only rows that match every filter on their side are compared. Filters apply to every Test in the Table Pair, and summaries and mismatch evidence are computed from the same filtered rows. Each Run reads the source and the target independently; there is no atomic snapshot across the two. Lakehouse reads use the current Delta version at the time of the read unless the Table Pair pins one in Source Delta version or Target Delta version; the placeholder Latest at Run time means no pin. The version actually read is recorded with the Run’s details in your Results Lakehouse. Warehouse Tables cannot be pinned. Column choices in the Test dialog always reflect the current Table, even when the pair pins an older version. A Table Pair’s source and target Tables are fixed once it is created; see Create Table Pairs.

Plan

Create Tests

Run and review

Common workflows

Supported inputs

The 12 validation functions

Limits

​Supported inputs

​The 12 validation functions

​Limits

Supported inputs

The 12 validation functions

Limits