I'm doing some planning around vCPU contention ratios, and I'm curious as to whether there is anything actually published by VMware. I've variously heard that the planning range should be 5 - 8, or 5 - 9 or 6 - 10 (e.g. here: http://communities.vmware.com/thread/315250), but are these ranges actually published by VMware? If so, where?
Additionally, when this 5 - 8 is quoted, is it on vCPU per physical core (i.e. not considering hyperthreading) or vCPU or logical core (considering hyperthreading).
Before I'm deluged with answers such as "it depends on workload", "high vCPU per VM could worsen performance", "do some performance analysis first", etc, I am aware of these considerations, but the brief I have does not faciliate such considerations, and hence the focus on vCPU per core.
Thanks in advance.