Essendon wrote: » You are mostly correct. How resources work in a pool are like this - VM1 usually uses say 16GB RAM and if suddenly wants say 32GB, the pool hands it out. Then when VM1 doesn't need it any more the RAM flows back in to the pool for other members of the pool. It's best to set shares inside resource pools so in times of contention VMs with higher shares get more. If there are say 10 Exchange servers that need serious grunt, it might be better to put them in a cluster of their own. Sure there's a slight increase in overhead but that way your dead certain no other VM will victimize your Exchange servers (at least compute-wise, you still gotta watch your storage and network). I didnt quite get what you meant by the last sentence in your post. HA is for a cluster, not a pool. HA doesnt starve other objects of resources, it is a mechanism to get your VM's going on another host when a host fails. The resultant effect is that the number of VM's on each remaining host increases and each VM then gets less resources to play with. Make sense?
Essendon wrote: » These are good questions, I suggest you read the resource pool section in the Mastering vSphere book and lab it up as you go. Things will be far clearer if you saw them happening yourself.