Deep Dive in Upgrading SQLServer Cluster with Distributed AlwaysOn Group

Stacey Kyrylova

The main purpose of this session is to review a SQL Server cluster upgrade that has Distributed AlwaysOn group (DistAG) for Geographic Read-Scale and transactional replication to support Read-Only requests from web services. We’ll also take the time to pause and evaluate the challenges a DBA will face during the upgrade of such a cluster and how to deal with them.

The session will start with a brief recap of what DistAG is and its purpose, and introduce a real-world setup of a SQL cluster where upstream primary DistAG not only supports Read-Write requests from ETL processes but also has transactional replication to another standalone SQL server.

The main part of the session will show what limitations and downtime a DBA should expect the system to experience during the DistAG upgrade when the general upgrade steps recommended by Microsoft are executed. Based on observation from the demonstrated upgrade, an alternative upgrade plan will be proposed and tested during the presentation. At the end of the session, participants will find out whether it is possible to have a rolling upgrade of DistAG with zero downtime or not.