I am starting to dabble in VM and Host affinity rules for the first time. One of our customers is looking to have 30-40 VMs spun up to use for distributed processing. They want to ensure that in case of a host failure, they will have as few VMs down at any time. It seemed that we should be able to use vm anti-affinity rules to meet that goal of forcing them to be spread evenly across all available hosts in the cluster.
I haven't had much luck tracking down example scenarios, so I had some questions on how it would work. I was thinking of creating a single VM Anti-affinity rule and putting all of the 30-40 VMs in there. I'm assuming, from what I've read, that it would do its best to satisfy the rule by balancing them out across all hosts, even though it might mean 2 VMs exist on the same host. Would this work? Would this cause any alerts or errors, if we had more VMs than hosts? Or would I need to setup multiple rules to allow this to work without error.
Would VM Affinity rules make the most sense here, or would host affinity rules be relevant here, and would one be preferential over the other?
Thank you.