I have some comprehension question concerning HA Resource Reservation. Let's assume I have a vSphere cluster with two nodes. Each nodes has 100 GB of RAM. So total RAM = 200 GB in my cluster.
I configure Admission Control with failover capacity 50% :
OK, now there is for example one VM with 20 GB of RAM and 100% RAM reservation. In addition there are 15 other VMs with 10 GB RAM each and they have all no reservation.
To keep the calculation simple let's assume there is no overhead Memory. So I think the Resource Reservation calculation is as follows:
Available Reservation: 200 GB
Used Reservation: 120 GB (50% from Admission Control and 20 GB from one VM)
Available Reservation: 80 GB
HA state of the cluster should be green, right?
Now a node failure happens. I have now 100 GB RAM left in the cluster. So the VM with the 20 GB Reservation is powered on - memory is guaranteed. All remaining 15 VMs with no reservations have to share the remaining Memory of 80 GB, so they compete for memory.
Is my assumption correct that Admission Control calculates the resources by considering the configured host failover capacity plus all memory reservations? And if there are enough resources to satisfy the needed amount of memory, the admission control state is okay?
Are the failover resources reported as okay even if there are plenty of other vms with no memory reservation as above-mentioned that have enough memory when the cluster is healthy but have to compete for memory in case of a host failure?
What would happen, if I configure the performance degradation VMs tolerate to 0 %... would there be a warning in case of a loss of 50% memory if the memory of each vm would become less than the memory allocated at the healthy state?
Will the reported Worst Case Allocation value for the VMs with no reservation getting lower if the resources are short in case of a host failure? (In my lab, every VM reports its configured amount of Memory as Worst Case Allocation.... so what must happen that the Worst Case Allocation Value is getting lower? (I know, if there are memory reservations, the worst case allocation will be at least the Reservation)
Performance degradation VMs tolerate feature issues a warning when a host failure would cause a reduction in VM performance based on actual resource consumption, not just the configured reservations. By setting this to 0%, you can effectively change Admission Control to work for configured memory as opposed to reserved memory, and therefore the performance and not just the availability of the VMs can be preserved during an event.
Thx for your answer.
What would be the value for the worst case Allocation in my scenario? For the vms with no reservation, is it calculated like this: 80 GB / 15 VMs = 5,3 GB worst case allocation? (15 vms are competing for 80 GB available (=non reserved) memory)