Why is high availability a challenge for virtual appliances in public cloud environments?

Public cloud providers prefer to charge by the VM$-$hour, and an idle VM incurs no charges.

Link$-$local multicast is not supported in most public cloud environments like in AWS VPCs$.$

The virtual machine ID for a highly available pair would need to be the same between the active and standby virtual machines.

Deploying two virtual machines with the same virtual IP address would cause an IP conflict on the network, similar to using the same virtual IP address on multiple devices in a physical network.