How Azure SQL Data Warehouse Works

Azure SQL Data Warehouse uses parallel processing computing technology to deliver high performance on large queries. Parallel processing is a computing technique that distributes processing workloads across multiple computers, called a cluster. The illustration shows a typical SQL Data Warehouse configuration.

Queries in the T-SQL language (top left) are input into the service as they would be with other SQL Server databases, by connecting through an application, management tool, or business intelligence application, like Excel or Power BI.

Queries are received by the control node (top), which parses queries, generates query plans, and distributes the workload to the compute nodes (center). Compute nodes access data warehouse data directly, which is stored in Azure Storage blobs (bottom), and return results to the control node for aggregation.

Each node includes an instance of the SQL Server database engine and a Data Movement Service (DMS), which is responsible for data transfer and partitioning operations, a key performance feature in parallel processing.