Provide a means to mark an IaaS Virtual Machine as unhealthy or 'not yet ready'
Currently when the infrastructure decides to reboot a VM for maintenance there is some kind of internal probing done by the Hypervisor to determine when the VM is healthy and then to reboot VM's in other fault/upgrade domains.
The conditions under which a VM is considered healthy is too optimistic.
Our database servers need 5-15 minutes to 'warm up' before being 'healthy'. This means the infrastructure is taking down both our primary and secondary virtual machines in a time frame that overlaps with both their warmup periods, bringing down our production application.