PolyBase features and limitations

Applies to: SQL Server 2016 (13.x) and later versions Azure SQL Database Azure Synapse Analytics Analytics Platform System (PDW)

This article is a summary of PolyBase features available for SQL Server products and services.

Feature summary for product releases

This table lists the key features for PolyBase and the products in which they're available.

Feature SQL Server (Beginning with 2016) Azure SQL Database Azure Synapse Analytics Parallel Data Warehouse
Query Hadoop data with Transact-SQL Yes No No Yes
Import data from Hadoop Yes No No Yes
Export data to Hadoop Yes No No Yes
Query, import from, export to Azure HDInsight No No No No
Push down query computations to Hadoop Yes No No Yes
Import data from Azure Blob storage Yes Yes 1 Yes Yes
Export data to Azure Blob storage Yes No Yes Yes
Import data from Azure Data Lake Store No No Yes No
Export data to Azure Data Lake Store No No Yes No
Run PolyBase queries from Microsoft BI tools Yes No Yes Yes

1 Introduced in SQL Server 2017 (14.x), see Examples of bulk access to data in Azure Blob Storage.

Known limitations

PolyBase has the following limitations:

  • Before SQL Server 2019 (15.x), the maximum possible row size, which includes the full length of variable length columns, can't exceed 32 KB in SQL Server or 1 MB in Azure Synapse Analytics. In SQL Server 2019 (15.x) and later versions, this limitation is lifted. The limit remains 1 MB for Hadoop data sources, but is limited only by the maximum SQL Server limit for other data sources.

  • When data is exported into an ORC file format from SQL Server or Azure Synapse Analytics, text-heavy columns might be limited. They can be limited to as few as 50 columns because of Java out-of-memory error messages. To work around this issue, export only a subset of the columns.

  • PolyBase can't connect to any Hadoop instance if Knox is enabled.

  • If you use Hive tables with transactional = true, PolyBase can't access the data in the Hive table's directory.

  • PolyBase services require SQL Server service to have TCP/IP network protocol enabled to function correctly. Additionally, if TCP/IP Protocol configuration setting Listen All is set to No, you must still have an entry for the correct listener port in either TCP Dynamic Ports or TCP Ports under IPAll in TCP/IP Properties. This is required due to the way PolyBase services resolve the listener port of the SQL Server Engine.

  • PolyBase on SQL Server on Linux doesn't function if IPv6 is disabled in the kernel. For more information, see SQL Server on Linux: Known issues.

  • PolyBase services require Shared Memory protocol to be enabled to function properly.

  • If you have a default SQL Server instance that is configured to listen on TCP port other than 1433, you can't use it as a head node in a PolyBase scale-out group. When you execute sp_polybase_join_group, if you pass 'MSSQLSERVER' as the instance name, SQL Server assumes port 1433 is the listener port, so the Data Movement service is unable to connect to the head node when starting.

  • Oracle synonyms aren't supported for usage with PolyBase.

  • UTF-8 collations aren't supported for Hadoop external data sources.

  • Hadoop is no longer supported in SQL Server 2022 (16.x) and later versions.

  • PolyBase scale-out groups are not supported with Always On availability groups.