Azure SQL Data Warehouse was designed to support a huge amount of big read operations and full scans. That's why the physical design of tables has three extra differences against common SQL Server databases:
- Storage:
- By default, every table stored in Azure SQL Data Warehouse is clustered columnstore index
- Clustered B-tree index is also supported, and it must be explicitly defined within table creation
- Heap is also supported, and it must be explicitly defined within table creation
- Distribution: As massively parallel technology, every table is distributed across all compute nodes. The distribution could be solved in two modes:
- Round-robin: Default mode of distribution. Control node distributes data randomly ...