In order to deploy a service on Azure virtual machines with the goal of maintaining service availability during a datacenter failure, which Azure feature should be incorporated into the virtual machine deployment strategy?