When the VM comes under load, can it utilize all of the processing power on the host(12x2.49GHz), or only the 2x2.49 GHz because it only has two vCPUs?
In that case a 2 vCPU VM can only utilize the processing power of 2.x2.49Ghz. A single vCPU is always mapped to a single physical core/thread at any given point time. The hypervisor does not stretch vCPUs across multiple physical threads/cores. This probably wouldn't make much sense due to cache and memory access issues or such.
I guess a follow up to the question is, would adding more vCPUs help with more processing power on the VM? I can only assume it must since I've seen high performance VMs with multiple procs, however, I thought that in some cases multiple procs increased %ready due to CPU scheduling.
Yes it surely would, the more vCPUs a VM has, the more physical resources it can potentially make use of. The %ready time issue typically occurs when you have (more or less significantly) more vCPUs than physical threads available (oversubscribing), and most of these vCPUs demand CPU cycles at the same time. While a thread is busy with scheduling a vCPU, the other vCPUs demanding cycles can't be scheduled and have to wait. This is what %RDY time basically is.
It really depends a lot on the workload. For CPU-intensive workloads you might want to aim for a 1:1 ratio of total vCPUs/physical threads of the host, but for things like virtual desktops where you have lots of systems and most of them aren't very busy all the time, ratios up to 20:1 are not that rare.
You can find really detailed information on the ESXi CPU scheduler in this awesome whitepaper:
http://www.vmware.com/resources/techresources/10345
--
http://alpacapowered.wordpress.com
