Hi all,
I'm trying to understand the concepts and details of resource pools in Hyper-V in Windows Server 2012. It seems as if there is almost no documentation on all that. Perhaps somebody can support me here, maybe I've not seen some docs yet.
So far, I learned that resource pools in their current implementation serve mainly for metering purposes. You can create pools per tenant and then group VM resources into those pools to facilitate resource metering per tenant. That is, you enable metering once per pool and get all the data necessary to bill that one customer for all their resources (without metering individual VMs). Is that correct?
Furthermore, it seems to me that an ethernet pool goes one step further by providing an abstraction level for virtual switches. As far as I've understood you can add multiple vSwitches to a pool and then connect a VM to the pool. Hyper-V then decides which actual switch to use. This may be handy in a multi-host environment if vSwitches on different hosts use different names although they connect to the same network. Is that correct?
So - talking about actually managing that stuff I've learned how to create a pool and how to add VHD locations and virtual switches to a pool. Enabling resource metering for a pool then collects usage data from all the resources inside that pool.
But now: I can create a pool for memory and a pool for CPU. But I cannot add resources to those. Neither can I add a complete VM to a pool. Now I'm launching a VM that belongs to a customer whose resources I'm metering. How will Hyper-V know that it's supposed to collect data on CPU and memory usage for that VM?
Am I missing something here? Or is pool-based metering only good for ethernet and VHD resources, and CPU and memory still need to be metered per VM?
Thanks for clarification,
Nils
Nils Kaczenski
MVP Directory Services
Hannover, Germany