🗄️ Azure Synapse SQL Pools¶

SQL Pools in Azure Synapse provide both serverless and dedicated SQL compute options for data warehousing and analytics workloads.

🌟 Overview¶

Azure Synapse SQL Pools offer two distinct compute models to meet different analytical needs:

Serverless SQL Pools: Pay-per-query model for ad-hoc analytics
Dedicated SQL Pools: Reserved capacity for consistent, high-performance workloads

Both share T-SQL compatibility and seamless integration with Azure data services.

📊 SQL Pool Types Comparison¶

Feature	Serverless SQL	Dedicated SQL
Pricing Model	Pay per TB processed	Reserved DWUs
Infrastructure	No management	Managed capacity
Use Case	Ad-hoc queries	Data warehousing
Scale	Automatic	Manual/Scheduled
Performance	Variable	Predictable
Storage	Data Lake	Internal + External
Startup Time	Instant	Minutes
Best For	Exploration, BI	Production DW

🎯 Pool Selection Guide¶

Choose Serverless SQL When¶

Querying data lake files (Parquet, CSV, JSON)
Exploratory data analysis
Cost optimization for variable workloads
No infrastructure management needed
Prototyping and development
Infrequent query patterns

Choose Dedicated SQL When¶

Consistent high-performance requirements
Enterprise data warehousing
Complex ETL/ELT workflows
Predictable workload patterns
Advanced performance tuning needed
Regulatory compliance requiring dedicated resources

🗂️ Pool Components¶

⚡ Serverless SQL Pools ¶

Query data in your data lake without provisioning infrastructure.

Key Features: - Automatic scaling - T-SQL support for file formats - No infrastructure to manage - Integration with data lake

📖 Complete Guide →

🏢 Dedicated SQL Pools ¶

Enterprise-grade data warehousing with predictable performance.

Key Features: - Massively Parallel Processing (MPP) - Advanced performance features - Enterprise security - Predictable performance

📖 Complete Guide →

🚀 Quick Start Examples¶

Serverless SQL: Query Data Lake¶

-- Query Parquet files directly
SELECT
    ProductID,
    ProductName,
    SUM(Quantity) as TotalQuantity,
    SUM(Revenue) as TotalRevenue
FROM OPENROWSET(
    BULK 'https://myaccount.dfs.core.windows.net/sales/year=2024/*/*.parquet',
    FORMAT = 'PARQUET'
) AS sales
GROUP BY ProductID, ProductName
ORDER BY TotalRevenue DESC;

Dedicated SQL: Create Warehouse Table¶

-- Create distributed table
CREATE TABLE dbo.FactSales
(
    SalesKey BIGINT NOT NULL,
    DateKey INT NOT NULL,
    CustomerKey INT NOT NULL,
    ProductKey INT NOT NULL,
    Quantity INT NOT NULL,
    Amount DECIMAL(19,4) NOT NULL
)
WITH
(
    DISTRIBUTION = HASH(CustomerKey),
    CLUSTERED COLUMNSTORE INDEX
);

Last Updated: 2025-01-28 Service Version: General Availability Documentation Status: Complete