|
There are a couple ways to do it but chargebacks are complex in general ... One is to charge based on the allocated size regardless of thin provisioning. This is the service provider model. If a user requests 2TB, you charge for 2TB even if they only consume 200GB to start. In this model, your goal is to make the array profitable and use some of the profits to fund the expansion of the array as you continue to sell it.
Another model is to determine the RAW cost per GB in your array and charge back against RAW consumption by business unit. If, for example, the cost of your array is 100,000$ and it has 10TB of RAW space, each RAW GB is 10$. If your accounting department requests a 2TB volume at RAID1, then you charge them $40,000. Or you can charge them based on actual usage. If they request 2TB and consume 200GB (with RAID1, that will be 400GB), then you charge them $4,000.
The more complex you make the chargeback model, the more difficult it will become. As well, using something like AO and multiple tiers (SSD, SAS, NL) of volumes makes it more difficult to report on consumption and valuation.
The simplest model and the one I would suggest is to ignore tiers (if you have them) and use the service provider model. Restrict a business units access to the performance of the array based on Priority Optimization. That removes the monthly reporting. The drawback is that it makes the array appear to be much more expensive as you can oversell it. So you have to weigh the benefits vs drawbacks on that model.
|
