For many enterprises, managing cloud spending is their top cloud challenge. How can you optimize the existing use of the cloud in your enterprise while reducing costs? FinOps may be the answer.
Not so long ago, data flowed into a company from a limited number of sources, but that is no longer the case. Today, both the amount and the sources of data have exploded. Processing and storing all that data is driving up cloud costs. So is the increase in SaaS products (a 2021 survey found that enterprises are using an average 110 SaaS products) and the processing to run them. Over-provisioned or unoptimized cloud resources put even more demands on cloud processing. Sudden and unexpected surges in cloud data workloads frequently lead to cost overruns that break budgets, derail analytics projects, and slow innovation.
Strong economic headwinds are driving companies to increase the return on their investment (ROI) in their cloud data projects, meaning cost now has joined performance and quality as an SLA factor. Cloud-based analytics projects need a means of predicting and controlling these costs. Luckily, there’s a new approach to taming cloud data costs — Data+FinOps — that blends the agility of data observability with the financial governance framework known as FinOps.
A good FinOps program lets companies control cloud spending while making the best use of resources. It drives measurable, achievable returns on investment, while allowing enterprises of any size to turn cloud usage-based pricing models to their advantage by only paying for what they use and only using what they need.
Cross-functional teams can use FinOps as a guide as they forecast, monitor, and account for the cloud resources needed for analytics projects. A FinOps framework enables a data team to implement and maintain the most efficient use of cloud-based applications along with underlying storage, computing, and network resources.
The FinOps Five
In my previous article I discussed how to explain the benefits of FinOps to the C-suite. Once you’ve obtained buy-in, it’s time to build the FinOps framework. To make the most of your FinOps program investments, be sure your framework satisfied the following five requirements:
1. Predict and measure
Cross-functional FinOps teams need the ability to predict what resources a cloud analytics project will consume. They then need to be able to measure and compare their actual consumption and associated costs to their budget forecast.
Consider how poor visibility into high cloud infrastructure spending could lead to costs growing overnight as the result of a sudden increase in volume or a one-line configuration change. A similar error caused a customer data platform (CDP) company to see gross margins be described as a “black eye on the company,” but FinOps allowed them to improve their gross margin by 20 percent in three months.
2. Assess the root cause
Consider the case of a global branded food company faced with a $70,000 bill for a single query on a cloud data analytics platform. A multinational e-commerce company discovered a similar issue when a developer found a query on Google Cloud BigQuery that would have resulted in a $1 million monthly bill.
When costs exceed (or seem likely to exceed) forecast, data teams need the ability to assess the root cause of the overrun. That’s often easier said than done because as data travels from producer through various transformation steps, it becomes increasingly unclear who owns that data. A strong FinOps program enables data teams to determine the who, what, where, when, and why project costs derailed.
3. Foster collaboration
FinOps teams depend on transparent communication and efficient handoffs between multiple roles. It’s true that cloud cost governance requires teamwork. A typical team includes a finance manager, FinOps manager, project manager, data engineer, CloudOps engineer, data analyst, and/or data scientist.
FinOps can’t be us versus them. Everyone must take ownership of the entire cost — it can’t be a top-down mandate. Everyone needs to buy into the program.
4. Guide human action
For FinOps to work, stakeholders and practitioners need guidance from data observability tools as they make decisions and act. However, it needs to be easy for users to do the right thing with prescriptive recommendations. Automated alerts can help FinOps and finance managers intervene when costs exceed thresholds for any aspect of the project (e.g., budget, size, time), and AI can be used to generate advice on how and what to change — whether it be settings or configurations or code — to optimize for costs.
5. Bring business impact into the equation
FinOps helps align cloud investments with business objectives to maximize an organization’s return on those investments. A FinOps approach helps data teams measure the value derived from new data products, applications, and ML models and quantify the associated cloud costs. Enterprises are increasingly turning to cloud unit economics for actionable insights.
Start by calculating your costs per customer or costs per product. For example, Target combines cloud provider data with business data to produce unit economics insights. This requires granular visibility into cloud service consumption to allocate costs. Fidelity Investments leverages visibility into account, service, and provider spending to power unit cost reporting that empowers their teams to make better cloud capacity and usage decisions.
Companies such as Adobe have been investing in AI/ML for a long time. To support the launch of new AI products, organizations are leveraging a FinOps approach to get insights into their rapidly expanding cloud data analytics and AI investments. Using the FinOps methodology, they have achieved ROI goals for faster product launches and higher margins.
Crawl, Walk, Run
If your company wants to tighten its belt while increasing the ROI from its cloud data projects, consider FinOps as a means of optimizing cloud costs. A FinOps framework’s strategic initiatives will help your data teams improve your company’s profitability and stability and better manage and optimize cloud data workloads — but only if your FinOps framework is implemented properly.
Start by ensuring that any FinOps program meets the requirements described above. Make sure you allow for a degree of flexibility so your data teams can adjust their tools and processes to meet corporate needs.
Remember, it’s fine to start slow. After all, you must learn to crawl before you can walk. The guidelines presented here will help you get your FinOps program up and running efficiently and effectively.