Primary Key & Unique Constraints in Azure Synapse Dedicated SQL Pool | NOT ENFORCED Explained | Azure Synapse Analytics Tutorial

Primary Key & Unique Constraints in Azure Synapse Dedicated SQL Pool | NOT ENFORCED Explained

🔍 Introduction

In Azure Synapse Dedicated SQL Pools, constraints such as PRIMARY KEY and UNIQUE are allowed — but with a twist: they are NOT ENFORCED. This means they serve as metadata hints for the query optimizer rather than enforcing data integrity rules like in traditional SQL Server.

📌 Key Points

• FOREIGN KEY constraints are not supported.
• PRIMARY KEY and UNIQUE constraints must include NOT ENFORCED.
• Duplicates are not prevented — you are responsible for ensuring data quality.
• These constraints can help the query optimizer generate better plans.

🛠️ Demo Overview

Step 1: Create Demo Table

CREATE TABLE dbo.Products (
    ProductID INT NOT NULL,
    ProductName NVARCHAR(100),
    Category NVARCHAR(50)
)
WITH (DISTRIBUTION = ROUND_ROBIN);

Step 2: Insert Duplicate Values

INSERT INTO dbo.Products VALUES (1, 'Laptop', 'Electronics');
INSERT INTO dbo.Products VALUES (2, 'Tablet', 'Electronics');
INSERT INTO dbo.Products VALUES (1, 'Monitor', 'Electronics'); -- Duplicate ProductID
INSERT INTO dbo.Products VALUES (3, 'Desk', 'Furniture');

Step 3: Add UNIQUE Constraint (NOT ENFORCED)

ALTER TABLE dbo.Products
ADD CONSTRAINT UQ_ProductID UNIQUE (ProductID) NOT ENFORCED;

⚠️ The constraint will be created even though duplicates exist.

Step 4: Remove Duplicate Manually

DELETE FROM dbo.Products WHERE ProductName = 'Monitor';

Step 5: Replace UNIQUE with PRIMARY KEY (also NOT ENFORCED)

ALTER TABLE dbo.Products
DROP CONSTRAINT UQ_ProductID;

ALTER TABLE dbo.Products
ADD CONSTRAINT PK_ProductID PRIMARY KEY NONCLUSTERED (ProductID) NOT ENFORCED;

💡 Why Use These Constraints?

• Helps Synapse generate optimized query execution plans.
• Useful for BI/reporting layers that assume uniqueness.
• Important for documentation and data governance.

🚫 Limitations

• Constraints won't stop bad data — no runtime enforcement.
• You must manually ensure data quality via queries or ELT logic.

📺 Watch the Video Tutorial

📚 Credit: Content created with the help of ChatGPT and Gemini.