Rethinking Chargeback in the Era of Clouds

Chargeback has been around since the early days of mainframes, and the coming of the cloud has made the process simpler in some instances and more complex in others.  

Chargeback is as much a psychological exercise as a financial one. When users have to pay for the IT resources they consume, they have incentive to be more responsible and efficient.

The easiest way to implement chargeback is to split the cost of IT across all departments or employees and charge everyone the same amount, but this defeats the awareness-raising purpose. More effective approaches include charging for time spent on the computer, the quantity of computing resources used, the number of active users or some combination of all those. However, this adds complexity for what is basically an accounting exercise.

Public cloud makes chargeback easier to administer. In the simplest case, users can sign up for cloud services with their credit cards and expense the cost. That might be fine for a small organization, but enterprises will want to take advantage of the discounts and management tools that cloud service providers offer their larger customers.

Fortunately, service providers are pretty good at taking chargeback off your hands. In the case of infrastructure as a service, you can assign each department, work group or individual a unique virtual machine and instruct the IaaS vendor to give you a breakdown of costs by computing instance. Be sure to consider the resources you allocate to each instance, because IaaS providers will charge you by the size of the virtual machine. Don’t allocate more CPU, memory and storage than you expect to need.

SaaS is even simpler because each user has a unique account. Most SaaS providers can easily provide you with an itemized bill broken down by user. Roll them up to the department level and you’re done.

The complexity arises when you get into charging back for private or hybrid clouds. That’s because virtual machines assign resources dynamically, and it can be a management chore to figure out who’s used what. If chargeback is important to you, consider using software provided by your virtualization software provider or constructing a simplified model that can be applied equally. A Google search will turn up several examples of models you can use.

But you might find that constructing a model isn’t worth the effort. A simple approach is to calculate the total cost of your computing resource and estimate the quantities consumed by each function or department. This can be done by assigning each group its own virtual machine instance and tracking usage over the period of one month. You can estimate resource consumption by any of the metrics described above. Allocate a percentage of the total cost to each department and bill back each month accordingly.

The problem with this approach is that it doesn’t reward users for efficiency. One way to address that is to conduct the same one-month exercise each year and re-adjust the percentages.

This also might be a good time to think about whether chargeback is really necessary. The IDG Enterprise 2014 Cloud Computing Study found that 23% of cloud spending now occurs outside of the IT department, a figure that will rise to 28% in the future. As IT increasingly migrates to a service model, organizations can offload accounting and focus on more important things. That’s good, because I can’t remember the last time I heard a CIO boast about his organization’s superior approach to cost accounting.