News

Microsoft Ending Support of SQL Server 2019 Big Data Clusters

• By Kurt Mackie
• 03/09/2022

Microsoft announced late last month that it'll be ending support for two SQL Server capabilities, with cloud services and SQL Server 2022 typically proposed as replacements.

The SQL Server 2019 Big Data Clusters add-on and the PolyBase Scale-Out Groups feature are getting dropped. Microsoft generally wants organizations to plan to switch to SQL Server 2022, which is still at the preview stage, and use some upcoming connector solutions, as a replacement.

The changes will affect SQL Server 2019 users, plus users of SQL Server 2017 and SQL Server 2016 as well with regard to the PolyBase Scale-Out Groups feature.

Big Data Clusters Ending March 2025
The SQL Server 2019 Big Data Clusters add-on will lose support after February 2025. However, Software Assurance is needed just to get that level of support for the Big Data Clusters add-on, per the announcement:

All existing users of SQL Server 2019 with Software Assurance will be fully supported on the platform for the next three years, through February 2025. This software [Big Data Clusters add-on] will continue to be maintained through SQL Server cumulative updates until that time.

No support details were described for organizations that might be running SQL Server 2019 without Software Assurance.

The Big Data Clusters add-on for SQL Server 2019 offers a way to "deploy scalable clusters of SQL Server, Spark, and HDFS [Hadoop Distributed File System] containers running on Kubernetes," according to Microsoft's description. The add-on was designed to ease deployment and management for those scalable clusters, per a Microsoft expert description.

If organizations using the Big Data Clusters add-on still want to conduct analytics using an on-premises solution, then Microsoft is suggesting organizations use SQL Server 2022, currently at the preview stage. SQL Server 2022 will also be able to integrate with cloud-based analytics services using Azure Synapse Link for SQL Server 2022, which "will be generally available to purchase later this year," the announcement indicated.

Alternatively, for organizations that can use cloud-based services, Microsoft is proposing a switch to using Microsoft Azure technologies, per its "Big Data Options" document. Those technologies include Azure Synapse, Azure Machine Learning and Azure Databricks, among others.

PolyBase Scale-Out Groups To End
Microsoft also plans to discontinue PolyBase Scale-Out Groups in the coming SQL Server 2022 product.

Older SQL Server products, such as SQL Server 2019, 2017 and 2016, will have support for PolyBase Scale-Out Groups throughout those server products' lifecycles.

Microsoft has described PolyBase Scale-Out Groups as "a group of SQL Server instances" that are used to improve "data loading and query performance." The PolyBase Scale-Out Groups feature permits the processing of "large external data sets in a parallel processing architecture," according to Microsoft's description.

Microsoft also plans to end support in SQL Server 2022 for Cloudera and Hortonworks external data sources. Organizations should switch to SQL Server 2022's "new object storage integration" instead.

Here's how it was described in the "Big Data Options" document:

Moving forward, support for external data sources will be limited to product versions in mainstream support by the respective vendor. You are advised encouraged to use the new object storage integration available in SQL Server 2022.

Migration Advice
Microsoft offered painful advice for organization eyeing the end of the SQL Server 2019 Big Data Clusters add-on and the PolyBase Scale-Out Groups feature. It can be a go hybrid or go cloud type of decision, per the "Big Data Options" document:

Once you pick a location (In-Cloud or Hybrid) for your migration, you should weigh the downtime and cost vectors to determine whether you will run a new system and move the data from the previous system to the new in real-time (side-by-side migration) or a backup and restore, or a new-start of the system from existing data sources (in-place migration).

Organizations could just change the connection strings for a migration, but they may also may have to rewrite code, the document explained.