Score:2

Windows Failover Cluster With SQL Server

es flag

We are moving away from using VMWare over to Windows Failover Cluster (version 2019) to host our company services. I have successfully configured WFC and have a few Virtual-Machine roles running along with the File-Server role and DHCP-Server role. Next is to get our SQL server instances (also version 2019) setup and I am posed with an architectural question.

Is it acceptable to install SQL server clustered role directly onto the bare-metal WFC hypervisors? Let me take a step back..

We want SQL server to be highly-available, just like anything else in the cluster. Following articles such as https://www.mssqltips.com/sqlservertip/6586/clustering-sql-server-2019-step-by-step/ it shows how to setup an always online SQL server instance, which uses WFC as a foundation to facilitate failover between the SQL server instances.

My thought is, why waste resources by firing up two new virtual machines, turning these two machines into a WFC, then installing SQL AO into that WFC when it can just be done in the parent instance? It seems overly redundant, the sql servers can failover within the virtual WFC and each vm can failover within the parent WFC. It seems like this was designed to be installed directly on the main WFC instance alongside other roles.

What is best practice here?

Score:4
gb flag

SQL Server running on top of VM in the cluster will be highly available. Failover Cluster with Hyper-V will handle VMs Failover. MS best practices states that Hyper-V should be the only role installed on a host.

https://docs.microsoft.com/en-us/windows-server/virtualization/hyper-v/best-practices-analyzer/hyper-v-should-be-the-only-enabled-role

You can deploy nested failover cluster though. Shared VHDX could be used to configure SQL Server Cluster.

https://docs.microsoft.com/en-us/previous-versions/windows/it-pro/windows-server-2012-r2-and-2012/dn265980(v=ws.11)

Check the following article to improve SQL FCI performance.

https://www.hyper-v.io/can-sql-server-failover-cluster-instance-run-twice-fast-sql-server-basic-availability-groups-2-node-cluster-part-2-studying-fci-performance/

KKlouzal avatar
es flag
Thanks for the microsoft documentation. I have seen that before and if we were hosting a standalone Hyper-V Server then by no means would we want to run anything other than virtual machines. A Windows Failover Cluster is something a little different though. Documentation talks about the negative impact being an increased amount of windows updates which would compound the amount of downtime during updates. WFC lets you fail roles over to an alternate node before applying updates virtually eliminating downtime regardless of how many updates are required.
Stuka avatar
gb flag
It is mostly recommendation. According to this thread https://social.technet.microsoft.com/Forums/lync/en-US/f67b4431-0d37-424c-81fb-72913d5c1237/mixed-cluster-roles?forum=winserverClustering It is mostly recommended because of performance reasons. You can still deploy it that way and it should be supported.
mangohost

Post an answer

Most people don’t grasp that asking a lot of questions unlocks learning and improves interpersonal bonding. In Alison’s studies, for example, though people could accurately recall how many questions had been asked in their conversations, they didn’t intuit the link between questions and liking. Across four studies, in which participants were engaged in conversations themselves or read transcripts of others’ conversations, people tended not to realize that question asking would influence—or had influenced—the level of amity between the conversationalists.