Memory capping in chargeback to reflect licensing for vspp

[PAckermann]

Hi,

we are a vspp-partner and have therefore a memory-capping at 24GB vRAM per VM. My question is, how can I configure this in vcenter chargeback, so that we can bill our customers correctly? At the moment chargeback accounts any memory configured to a VM, so customers with a VM of say 32GB will be billed for 32GB vRAM instead of 24GB, what would be correct.

[IamTHEvilONE]

In Chargeback, you would look at the memory Size ... which is the vRAM allocation for a given VM.  It takes the value directly from vCenter, so whatever is configured there.

If you are capped to 24GB of vRAM per VM ... then how would you get above that on your VSPP License?  I mean, you state you have a restriction ... so is that a restriction of what you can enter into the VM's parameters?  I guess I'm wondering how the restriction is enforced.

Best Regards,

Jon Hemming

[IamTHEvilONE]

Otherwise, you could use allocation units ... e.g. set ram allocation to 24 but that requires manual intervention when the VM goes above/below the 24 gb threshold.

[PAckermann]

It is not a restriction. In VSPP licensing is per usage of vRAM, the capping kicks in, when a VM has more than 24GB of vRAM. To lighten this, here an example:

VM with 16GB vRAM will be licensed for 16GB

VM with 24GB vRAM will be licensed for 24GB

VM with 32GB vRAM will be licensed for 24GB

So the maximum of vRAM that should be charged is 24GB vRAM, even if the VM is configured with more.

We are a VMWare service provider and have some customers with dedicated VMWare-clusters as well as vCloud-services, where the customer can configure the VMs by them selfes. Most of our VMs running have less than 24GB vRAM, so no problem here, but we have now a customer who runs some VMs with 32GB and 64GB vRAM. Charging him the full vRAM would mean, that he would have near double licensing-costs. So I'm looking for a way to charge the VMs properly. If there is no built-in setting to do so in chargeback, a way through the API would be also ok.

[matt232]

Don't you still have to account for the cost of the RAM (which is likely more expensive due to the need for a larger DIMM size to meet the requirements), electricity and cooling that is being used? Or the opportunity cost of those extra 18-40GB that some other client is now not able to use?

[PAckermann]

For dedicated environments the customer gets billed for the hardware, including electricity, cooling, ... + VMWare-licenses. For this particular case I need a correct billing as I stated before. For a shared environment, where hardware-costs must be included you are perfectly right, but that is not the problem here.

This particular customer hosts 8 VMWare-nodes, each of 96GB RAM. For hosting these 8 nodes and the storage we charge him an amount X. What we need additionaly is charging for the VMWare-licenses. If we would charge only by the physical RAM, that would be 7x96GB (one failoverhost excluded) = 672GB vRAM! As the licenses are charged only by vRAM-usage, we won't charge the full 672GB. As charging for every GB in a VM gets very expensive for large VMs, it would be completely uninteresting for large companies, that was something VMWare realized after several complaints from hosters and introduced the 24GB vRAM capping.

To get an idea of the dicrepancy of what chargeback would report now and what should be correct:

4 VMs with 64GB vRAM

4 VMs with 32GB vRAM

chargeback would charge 4x 64GB + 4x 32GB = 384GB vRAM

correct would be 8x 24GB = 192GB vRAM

now say we would charge 4$ per GB vRAM. Chargeback would charge 1536$ per month, but the correct value would be only 768$ per month. Count some more VMs in and you get the idea.

[peterdabr]

We are facing similar problem with not being able to configure Chargeback to account for 24GB cap and reserved vRAM model.

One would think that Chargeback would provide such capabilities out-of-the-box to accommodate to a growing number of VSPP partners.

Does anybody know how to accomplish that without turning to API?

[zpeewee]

Hello,

Which kind of billing policy are you using ?

[PAckermann]

At the moment we are using "Maximum of Memory Usage and Memory Reservation", but as stated before, this does actualy not reflect the billing policy we need.

[zpeewee]

HOw are you sure that the value reported by chargeback can be use to calculate the VSPP point.

The actual usage is based on the metric memory consumed which is the (memory granted - memory shared).

Depending on your environnement, the actual usage may be lower than the memory defined into your virtual machine.

Or may be you are using reserved memory equal the memory size of the defined VM ?

[PAckermann]

I know, that the actual usage can differ from the provisioned, that is not the problem here. The problem is, when I have a VM that is provisioned with more than 24GB vRAM, it should only be charged for a max. of 24GB vRAM and thats exactly what I can't find in chargeback how to configure this.

[admin]

Can you send an email to [e-mail] and I can give you some more details on how the cap works.