The key to availability is redundancy. This means that the servers running Hyper-V must be resilient to failure. Core components such as power should not have single points of failure. Specifics around storage and networking will be discussed in their upcoming sections.
Examples include having more than one physical network adapter (NIC or pNIC) and/or Host Bus Adapter (HBA). NICs or HBAs often have multiple connections on a single card which are technically redundant, but if the card or backplane interface itself fails, you will lose connectivity to networking or storage. Consider options such as having multiple pNICs or HBAs on different hardware backplane connections so that adapter active multipathing will absorb a failure and either prevent or minimize an outage (assuming proper configuration).
At the Windows Server layer, should one hypervisor host fail, another must be able to now run the VMs from the failed host. In Windows Server, a Windows Server Failover Cluster (WSFC), enabled by the Failover Clustering feature, provides this functionality. For more information on WSFCs, see the Microsoft Learn article “Failover Clustering in Windows Server and Azure Stack HCI”.
ActiveCluster and ActiveDR can be important features of FlashArray to explore to protect both Hyper-V hosts as well as VMs. See the paper Microsoft Hyper-V Stretched Cluster with ActiveCluster for more information on Hyper-V and ActiveCluster. See the paper ActiveDR and Microsoft Solutions for more information on Hyper-V and ActiveDR.