So I assume then, if it is indeed partly a calculation of VM reservation, then this is why I am getting very different figures when looking at the cluster resource allocation figures. What I mean is that we have vCloud Director managing our platform and vCenter servers, we use the allocation model, which means reservation are set at the resource pool level and not at the individual VM level, hence why the Cluster Memory Reserved Capacity is way off as it doesn't include resource pool reservations.