Why should we be concerned with the showing of costs or the charging back of costs within internal IT departments?
Typically IT departments have been funded through an allocation of their budget costs to the business lines which they support. This was done as IT was considered an expense. Resources to run the organization were funded through this budget allocation or through a Capital Expenditure process and each resource was typically depreciated over a 3–5 year process. This process worked fairly well for the last 40 years or so.
However, with the advent of on-demand based resources, like Amazon and Salesforce, the typical funding model described above has led organizations to look for a different funding model. Instead of purchasing a physical server or a piece of software, there is now this concept of subscriptions to procure the resources necessary to support the organization’s IT needs. How does an organization move from a a funding model that is heavy on the Capital Expenditure model for purchasing new hardware to a model where we are looking at more Operational Expenditures that are not always known upfront when the service is implemented.
What is driving the change in resource consumption?
We only need to look at our own consumption of services. 40 years ago, you might have had a phone in your house and paid a monthly usage to the phone company for the right to use the phone. This and your electricity/gas utility were the only “services” that you would have probably worried about. Move this forward to the current century and we see that most things we utilize are priced as a “service.” Instead of purchasing movies, I subscribe to Netflix and pay for the ability to watch the movies over the Internet. I pay for Spotify to have access to music, instead of purchasing the music. I pay for cloud services, instead of additional hard drives. All of these are mothly charges that I have personally for services.
Organizations are moving to this model also. Microsoft offers Office365 for organizations that allow them to pay a periodic usage charge for the Office products. No longer is there an upgrade cycle with a huge outlay of costs to upgrade Office, but rather a periodic charge. This changes the budgeting model that an organization has to follow. Instead of budgeting for a large upgrade, they can now pass that usage charge back to the individual user’s cost center. By passing this charge back to the user’s cost center, it has the side effect to force the business lines to look at who is using the service and whether they have a need for the service. Out of this comes a potential cost savings as it no longer requires them to have the software available for all users, but only the users that need the software.
On an infrastructure side, the ability to pass the costs back directly to the business line can lead to more efficiencies. When you need to look 3–5 years out for the procurement of infrastucture, you tend to over purchase as you want to make sure you do not have to go back to the “well” and ask for more money. This has led a lot of organizations to have a large number of underutilized resources. To combat some of this underutilization, we have virtualized the resources internally, so that the overall utilization of the physical resources serving the virtualized resources looks higher. However, there is a still inherent wastefulness of resources that comes with that. Moving to a cost model starts to show the individual groups what those resources are costing them. This coupled with the showing of utilization metrics of those resources starts to paint the picture of how much it really is costing and also shows what wasted resources they are using.
Moving to a chargeback model.
The next step from the showing of costs is moving to a chargeback model. You start first by showing the costs, so that proper budgets can be created to account for the “real” costs of supporting the IT services. Then the next step is to allocate those costs to the business lines to go against those budgets.
What needs to be in place to move to this chargeback model? There has to be a single point of entry for the consumption of services. In most IT organizations there is a single point to “request” a resource, but there is a not necessarily a single point for the allocation of that resource. This is an important part of a chargeback model. This single point of entry allows for an approval process of that resource allocation and the assignment of that resource to a specific cost center. This is tracked in a database, so there is a record of each request and the disposition of that request.
Next, there needs to be a method for the tracking of the consumption of that resources. This can be as simple as just looking at the allocated resources (i.e. CPUs, GB Memory, TB Disk) or more complex, looking at the actual usage of those allocated resources. No matter which method is chosen, the consumption needs to be tracked at a granularity that makes sense to the organization. Most organizations can be content with an hourly or daily model, whereas some will want to get down to a minute granularity. This can even be different between “services” that are available. For example, we may charge container usage on a minute basis, but infrastructure on an allocation by hour. However, it is done, it needs to be consistent across the “service” being provided. There should not be different methods for costing infrastructure depending on who or what is utilizing it, this will just add to confusion.
Next will be the determination of what the costs are. These are your rate tables. These can be different depending on a number of criteria. You may charge a higher amount for CPU usage for a database server than you would for an application server, because the placement of that resource is into an environment that has a higher actual cost to the organization. These rates should be presented to the user when they procure a resource or service. They should be published and available, updated on a regular basis, so that all users know what the services will cost to them.
You may show back to users a different cost than the underlying provider of the resource. You may know that there is a standard cost for a virtual machine within your on-prem virtualized environment, but that is not the real cost, just the cost of the hardware. Your real cost to the user will be a cost of the hardware, software licensing, utility and support. This is the cost that you should show the user and charge back to their cost center. Some organizations will also give a discount to a group that leverages a certain provider or agrees to have their machines quiesced for a certain period of time each day. This may be especially true with development environments where there is not a need for 24×7 availability.
Conclusion
In conclusion, implementing a showback/chargeback model will not be without technical and political challenges. Some groups will not want to see the “real” cost of utilizing IT resources because they know it is more than the current allocation of charges they are seeing today. Other groups will not want to change, as budgeting becomes more involved since you will need to know what you need from a resource standpoint, instead of just purchasing more than you need.
Overall the implementation of this model should help gain efficiencies and potential cost savings across the organization. The cost of IT will more reflect the actual cost of operating the infrastucture to support the organization.
Periodic Ramblings 