Cluster Nodes and Availability Zones
A Hyper-V failover cluster is a group of servers, or nodes, that work together to keep virtual machines (VMs) running. To protect against a failure in a whole data center, you should place these cluster nodes in different Azure availability zones. This means if one zone has a problem, the VMs can automatically restart on a healthy node in another zone. For the best performance, you can use premium-managed disks for storage and organize related VMs into availability sets to further isolate potential failures.
Quorum Models
The cluster uses a quorum to decide which nodes are in charge and to prevent a "split-brain" scenario where two parts of the cluster try to run the same VM. The quorum is a majority vote, and for a two-node cluster, you must add a third voting resource called a witness. You can choose between a cloud witness (which uses an Azure storage account), a disk witness (which uses an Azure Shared Disk), or a file share witness. The cloud witness is often the best choice for clusters spread across different Azure regions.
Shared and Witness Storage
For the cluster to work, the nodes need shared storage that they can all access. Azure Shared Disks are designed for this purpose and support failover clustering. The witness resource also needs storage. If you use a cloud witness, the cluster nodes need outbound internet access on port 443 to talk to Azure Storage. If you use a file share witness, you must create an SMB file share with enough free space and ensure all nodes can connect to it over the network.
Validate Cluster Health
Keeping the cluster healthy requires regular checks. You must monitor the network connection between all nodes and the witness. It's also important to verify that each node's vote is correctly assigned, especially if a node or witness goes offline, to maintain the proper quorum. By running these validations, you ensure that the automated failover process will work correctly when needed, keeping your VMs available.