A cloud provider’s bill of cloud charges by the second for actual usage. With an even-spread cost model for cloud spend, decisions by one business unit impact the whole enterprise. Even-spread allocations actively incentivize poor cloud consumption decisions. When consumers aren’t accountable for their cloud spend there are no market forces to keep consumption in check— and everyone pays for it. Literally.
In this article, we explore bill of cloud, the challenges of cloud cost recovery, and unpack cloud cost allocation strategies.
Indirect costs of public cloud (physical labor, software, security, and other non-provider costs required to deploy and maintain services) aren’t in a cloud provider’s bill—and neither is data about consumption mapped to a business unit. A bill of IT for public cloud must include fully burdened cloud costs (direct and indirect) and defensible allocations of those costs based on BU consumption. A public cloud provider’s bill is a starting point to build a chargeback process, but it isn’t enough to build chargeback bill of IT for public cloud.
Public cloud providers position their value around agility and choice—business users need a cloud chargeback system that delivers on that promise. It’s impossible to compare costs in a multi-cloud world when every cloud provider has different service offerings and billing structures. A singular provider view of public cloud (AWS Cost Explorer or Azure Cost Management) doesn’t serve the needs of multi-cloud organizations.
We use showback to hold end-users accountable for their cloud usage. It helps end-users make better decisions. Our goal is to make this showback more holistic and include the true cost of operating an application, including the cost of services, networking, and labor.
Using chargeback to accelerate workload migrations (e.g., subsidize Amazon EC2 to accelerate the retirement of legacy hardware) depends on accurate pricing of the on-premises infrastructure you are moving from and the IaaS you are moving to. A revenue-neutral IT org risks slipping from cost recovery to profit center if they cannot compare the fully-burdened cost of the public cloud and existing on-premises options. This comparison is more nuanced once you bring in enterprise discount programs and variable pricing (e.g. spot vs. on-demand vs. reserved instances).
I&O leaders struggle to build defensible migration plans because they cannot compare between cloud providers and existing on-premises options. Instead, decisions are made based only on cloud provider pricing without a view of related support costs such as labor. Consequently, companies migrate their apps to the cloud (lift-and-shift, re-architect) but have no way to gauge whether they’ve achieved their migration targets.»Read more: 5 priorities for an effective cloud migration strategy.
Similar operational decisions in on-premises and cloud drive different financial outcomes. A depreciating server array has a fixed cost that doesn’t change based on usage. If it’s used or not used, needed or not needed: the depreciation schedule is locked and the financial commitment is the same. This isn’t the case for bill of cloud. Every server you have provisioned (used or not) is captured in your cloud provider bill: per second billing has consequences. The business needs to understand the billing implications of their behavior.
Good tagging is the key to avoiding paying for someone else’s cloud. Not all cloud users are good corporate citizens. An even spread allocation of cloud spend ties everyone to the consequences of (someone else’s) poor cloud cost management—the very definition of “paying for someone else’s cloud.”
Through 2020, 80% of organizations will initially overshoot their cloud infrastructure as a service budget, due to a lack of optimization governance or misguided upfront cloud spend commitments.
Gartner 10 Best Practices for Azure Cloud Cost Optimization Approaches to Allocate Cloud Costs
The relative simplicity of public cloud chargeback vs. on-premises chargeback drives unintended consequences. Pick the “wrong” type of cloud cost allocation and you will drive consumer behavior counter to your strategic goals.
There are four approaches to cloud cost recovery from your business partners. Cost recovery can be deemed fair (or unfair) with any one of these—“fairness” usually being a subjective assessment based on whether a stakeholder thinks the chosen cost allocation makes them a winner or a loser.
This is the quickest and easiest way to do cost allocation—even allocation of costs between business units. This method makes sense for any company in the early days of scrutinizing cloud spend and wants to quickly establish a baseline starting point.
An even spread produces a fast and accurate grand total but not necessarily fair individual burdening. Historically, organizations approached cloud spend as a skunkworks after-thought—the spend was too small to spend too much time worrying about fair cost recovery. Even spread allocations were good enough.
Cloud adoption is now so pervasive that benign neglect of fairness is no longer appropriate. In lieu of having nowhere else to start, even spread is the “foot in the door” of cost recovery—a place to mature from rather than to aspire to stay with.
Manually assigned percentage allocations provide more accurate cost assignments than even spread. This approach makes use of in-house knowledge of cloud costs to business units and takes advantage of existing spreadsheet-based financial models.
This approach makes sense when there is an IT finance role that understands how cloud allocation spend relates to different business units. However, the figures are only as good as the expertise of the person making the estimates.
In a manually weighted system, numbers represent consumption or activity. It puts control of costs in the hands of business users. A company uses this method when it makes more sense to directly assign the actual usage or consumption of resources than it is to calculate/estimate a percentage. This presupposes actual utilization data is available—not always a given.
Consumption-based allocations are the most mature, accurate and fair analysis of all these methods. This strategy tracks actual cloud activity captured in a system of record and then uses those numbers to distribute shared costs.
Consumption-based allocations rely on some of the methods employed in the cost-based and manually weighted strategies. It also incents behavior changes and puts control in the hands of cloud consumers. Help desks, PPM, and asset management systems are good sources for consumption-based allocations. Consider this approach if you are concerned with fairness, interested in improvement and change, and can collect usage data. Consumption-based allocations work for companies focused on accuracy because each line of business is assured that cloud costs are divided fairly—with no estimating or guessing.
Collecting usage data, and regularly updating it, is the biggest challenge in implementing this strategy. If a business unit changes its behaviors month-to-month, this may result in swings or shifts in allocation percentages.
»Read more: Stop paying for someone else's cloud
The right cloud cost allocations make IT and the business partners joint custodians of your organization cloud spend. The wrong allocation unearths latent hostility from consumers (business), provisioners (I&O) and purchasers (IT Finance). When choosing a strategy for moving costs from one cost pool to another, you must weigh several factors:
These cloud cost allocation methods offer different options on how to best share responsibility among business units and move toward the right efficiency and utilization targets. Regardless of your goals and cultures, the range of strategies detailed here assures that every enterprise adopting TBM will find the right fit.
»Download these eBooks to optimize hybrid and cloud by right-sizing legacy spend, shifting savings to innovation and efficiently embracing cloud: