If you have not yet implemented a chargeback in your organization, you are lagging behind in the IT industry. Almost every IT department has one today. Yet, not everyone uses it for the same reasons. Chargeback systems were originally created to help recuperate the IT budget from its customers, the business units. With the adoption of Service Management, IT departments used chargeback systems to show that they are focused on the customer and have some financial accountability. Some even used it to protect against outsourcing by demonstrating a lower or at least competitive cost of implementing and managing the IT infrastructure. While all these are valid reasons to use chargeback systems and should be thoroughly considered, one of the fundamental objectives of an IT chargeback system is to induce the right customer behaviors within a company.
A mature chargeback systems will represent the unit cost estimate, that of a server processor, storage gigabyte, network bandwidth in mbps, etc. A unit cost ensures that the customers will strive to use only as much as needed and no more thus minimizing waste and extraneous costs. It allocates the IT costs equitably to all of its consumers and creates a feeling of fairness and true representation of the cost of doing business.
Arriving at the unit cost may be a bit complicated as at the end of the day, IT chargeback should be a zero sum system: IT consumers pay for all the costs of IT, including hardware and software, personnel and their training, engineering labs and testing of new technologies. Yet, done right, the total cost of building and managing the IT infrastructure would be lower (approximately 40% lower) of that of an outsourced provider since it does not include the outsourcing vendor profit margin.
The need to transparently include all IT costs in a chargeback system needs to be balanced by the types of chargebacks a customer pays for. Too many chargeback types will create a financial nightmare and too few will confuse your customers about what they pay for and more importantly what to do to optimize their costs (what needs to be reduced or optimized). It is important to identify the units that are driven by business requirements and incorporate all adjunct charges in them (e.g rack space, power consumption, cooling, etc.) Ultimately, these charges should be hidden and not itemized for the consumer of IT services. Consider the primary contributors of the infrastructure costs and ask:
“Can the customer control the cost of this unit directly or is it the responsibility of the IT department to optimize that cost?”
Take, for example, power. While power consumption is a significant cost of IT infrastructure, its cost optimization is the sole responsibility of IT. By choosing the right hardware, an IT department can reduce the power draw. Implementing server and other technology features can reduce power consumption as well. On the other hand, the responsibility of optimizing storage capacity (gigabytes of storage) lies with the consumer, though some of it still remains with the storage department. The total cost per gigabyte of storage used per month per tier of service should include the related costs such as the power and cooling needed, the people to perform the required proactive and reactive tasks, the floor-space for the storage frame, storage area network switches, inter-switch cabling, etc. In order to get that cost, you should, of course, perform a complete Total Cost of Ownership (TCO) analysis to determine all cost components, variable and fixed, that go into supporting that storage capacity. Putting the responsibility back on the consumer, will ensure that the customer optimizes the total ask for storage capacity or at least they are aware how they can reduce the costs by reducing capacity and types of storage they use.
Listing the individual unit costs will, in many cases, be proven insufficient to influence the right behaviors. That is why the price (chargeback) of a unit needs not be directly proportional to its total cost of ownership. For example, you may find that the organization is inadvertently conservative and paying 80% more for tier I storage is not sufficient deterrent to choosing something they may not need (after all, most technologists want the best they can get their hands on.) That is why, you should at least employ these two techniques when you are building your chargeback system.
First, price higher what you want your consumers to be more diligent with. If everyone chooses for example Tier I service because the organization is super-conservative and less sensitive to the value/cost trade-off, then you should price Tier I higher than its proportional TCO. For example, price your Tier I primary storage at $2.50/GB/month even if the actual cost is $1.80/GB/month. If your Tier II primary storage is at $1/GB/month, the difference now between Tier I and II is significant to give the customer a pause and at least some more thought to it. Similarly, if everyone is very cost sensitive and tend to choose Tier III primary storage rather than appropriately Tier II, lower the Tier II storage and increase Tier III storage to reduce the difference and help drive the applications that need Tier II to the correct service tier.
Second, help your customers choose the right service tier by demonstrating the “should-be versus will-be” costs for their chosen services. If you can measure the requirements (e.g. performance) then you can clearly demonstrate that their applications can use Tier II versus Tier I or Tier III instead of Tier II. More importantly, if you can show them that an application using 2TB per month can result in a chargeback difference as high as $9000 per month (including OR and DR copies) then it makes it more tangible than focusing on the unit difference at $1.50 per GB. You should always provide this perspective particularly when the requirements are not as discrete. Choosing OR (Operational Recovery) or DR service tiers is usually subjective to the customer. Showing that OR I with near zero RTO and RPO costs several thousands of dollars but OR II with a couple of hours of RTO and RPO may only cost a fraction of that, will, more often than not, help the IT customer make the right decision (which is not always choosing the lower tier).
Your Chargeback model should follow closely your IT service catalog. Each service area will have respective unit charges based on the service tier and the # of units consumed. Your service areas are driven by business requirements and can be at a minimum the following ones:
- Primary Compute (unit price can be by type and number of servers, # of cores or mhz or combination of each)
- Primary Storage (unit price should be gigabytes by service tier, or when auto-tiering is a reality, % of max storage by storage tier)
- Operational Recovery (will calculate based on both primary compute and primary storage and the overall solution in place, note: this keeps your OR solution for storage and server consistent)
- Disaster Recovery (similar to OR, though it may have different charges than OR due to different solutions and adjunct costs)
- Archival Management (will be determined by GB consumed in the long term archiving platform, by archiving platform)
- Security Classification (will calculate based on the additional infrastructure that will be required per security service tier)
While the network (both LAN and SAN) is a major and critical infrastructure area, unless you are are able to measure the network utilization by application, you should allocate its costs in the other service areas. For example, an application that requires Primary Compute Tier I and Primary Storage Tier I is likely to use more bandwidth (and IOs) on both LAN and SAN and therefore it will bear a higher percentage of the network cost. An application that uses (a)synchonous replication across two data centers will bear that respective charge as part of a higher DR solution.
Chargeback is about recovering the IT costs in an equitable way. It is more about demonstrating IT improvements through lower unit charges year after year. It is much more about influencing and guiding the IT customers to properly protect their applications while making good use of the corporate investment and optimizing infrastructure costs. After all, it helps our IT customers be as efficient and effective as they expect us to be!
