Never Miss a Beat: Get a snapshot of the issues affecting the IT industry straight to your inbox.
Wisdom From the Pump: What Gas Spending Can Teach Us About Managing Public Cloud Costs
Both gas and cloud spending are on the rise, but there are common-sense efficiency steps organizations can take to get cloud spending under control.
4 Min Read
Elena Akifeva via Alamy Stock
Gas prices -- and driving miles -- are on the rise. Americans spend about 3.2% of their budget on gas, and cumulatively spend over $400 billion at the pump every year.
There’s a parallel in cloud computing, where the numbers are even scarier. Cloud spending, a top-three budget item for digital businesses, is projected to reach $679 billion this year, and exceed $1 trillion by 2027. Runaway spending -- not to mention the rising tide of generative AI -- means rising pressure on businesses to get cloud costs under control.
Fortunately, cloud financial operations (FinOps) was conceived for companies facing this pressure. Here are some common-sense steps organizations can take to get cloud spending under control, guided by principles that take root in the fuel efficiency world:
1. Give the right people the right data.
Efficiency depends on data quality. The better you understand your tools, your desired outcomes, and the results of past initiatives, the more you can hone your approach.
In a fuel context, “the right people” means anyone who drives. The more fuel efficiency data drivers have, the more fuel-efficiently they’ll drive. Fuel efficiency data would include things like the car(s) you have, the types of trips you tend to take, and the vehicles’ fuel efficiency ratings.
Cloud efficiency data would include the cost of the cloud resources your organization pays for, the sizes of those resources, and dimension costs -- i.e., the costs of individual customers, products, features, teams, microservices, etc.
In a cloud context, “the right people” would be anyone consuming cloud resources -- i.e., your engineers. The more cloud efficiency data your engineers have in a context that’s relevant to them, the more cloud-efficiently they’ll build.
2. Use the right vehicle for the job.
I have two very different Toyotas: A Sequoia that averages 15 MPG, and a RAV4 that averages 30 MPG. At $3.50/gal, that means my Sequoia gets 4.29 miles-per-dollar (MPD), and my RAV4 gets 8.57 MPD. Given this disparity, the rule for my household should be: Use the RAV4 unless you absolutely have to use the Sequoia.
The same principle applies in the cloud, but at much greater complexity. With vehicles, my household has only two options. In the cloud, the “vehicles” (instances) you could choose are far more numerous. Amazon EC2 alone offers more than 750 instance types.
Plus, many more “drivers” could be using those vehicles. In my company, I’ve got to make sure that dozens of engineers understand the efficiency data that’s relevant to them. Your organization might have hundreds or thousands of “drivers.”
The magnitude of the challenge tempts many organizations to find a cloud cost “easy button” and stop there. True, automated platforms can help organizations drive greater efficiency. But to get maximum cloud efficiency, you must combine automation with human decision-making.
For example, a resource that was once cloud-efficient may become inefficient as customers use the product/feature/microservice it’s powering. If you’re a product-led growth company whose free feature is getting used more heavily than you expected, its cloud costs will rise, and you’ll want to interrogate the underlying architecture to see if refactoring would increase efficiency. (Plus, you’ll want your engineers to be alerted in real time if the feature’s costs start to mount unhealthily.)
3. Maximize your efficiency rate.
The goal of using any tool is to use it efficiently: to maximize the return on your investment. In the case of cars, you’re not going to get much back financially, but you can consider metrics like “trips completed” or “miles driven” as “return on fuel investment.”
In the cloud, efficiency is a bit easier to quantify. For this, I recommend cloud efficiency rate (CER), which is calculated like this: Cloud Efficiency Rate = (Revenue - Cloud Costs) / Revenue.
Depending on your tooling, you can calculate overall CER and per-unit CER (CER per customer, per product, per transaction, etc.). The more units whose CER you can quantify, the more you’ll be able to see what drives efficiency, document efficient engineering practices, refine customer contracts, and more.
Consider the Impacts of Your Decisions at Scale
My family is fortunate that paying an extra $20/week at the pump isn’t going to spoil our weekend adventures, but that doesn’t mean our decisions are immaterial. For our two-car family, the impact of these decisions is in the thousands of dollars per year.
Now, consider a police department with hundreds of vehicles -- or Amazon, with nearly 300,000 delivery vehicles. Fuel waste left unchecked quickly becomes material, whether impacting the funds a local government can invest in services or impacting corporate profits and valuations. Now, imagine the efficiency gains possible against $1 trillion of cloud spend.
At Scale, Efficiency Decisions Make an Enormous impact
With investors showing renewed interest in profitability, organizations are facing renewed pressure to create sustainable cost models and improve their unit economics. Cloud spend continues to be a top-three budgetary concern and given cloud usage trends, will only become more severe in the years ahead.
About the Author
You May Also Like
More Insights
