Reading and Writing Parquet Files in SQL Server


 

When analytic data is stored directly in SQL Server, a clustered columnstore index is an ideal place for that data to reside. It is column-based, highly compressed, and supports a wide variety of optimizations for both reads and writes.

When analytic data is stored in files, such as for use in Azure Data Factory, Synapse, or Fabric, the Parquet file format will often be used. There are many similarities between these two columnstore formats and the methods they use to improve read/write speeds and storage space.

When writing data from SQL Server to files, though, uncompressed formats such as