Cloud consumption is growing rapidly, but tracking spending can be a bear. Enter FinOps, a financial discipline and tech solution intended to optimize how enterprise spend on cloud.
Won over by the promise to drive business agility while reducing costs, CIOs are migrating to the cloud en masse. But renting cloud services creates new problems, including managing bills populated with thousands of line items generated by instances running across the globe.
Increasingly, enterprises are turning to FinOps (financial operations), a business management discipline and analytics software designed to calculate the costs of public cloud services provided by vendors such Amazon Web Services (AWS) and Microsoft Azure. The approach aims to help enterprises better plan, budget and forecast spending requirements for cloud consumption.
Yet within this market, there remains a lot of waste, both of precious IT budgets and resource utilization. Eighty percent of 300 financial and IT leaders surveyed by 451 Research say that poor cloud financial management has had a negative impact on their business; 57 percent of respondents worry daily about cloud cost management, while 69 percent regularly overspend their cloud budget by 25 percent.
The economics of the cloud
More than a technology phenomenon for the 21st century, cloud is an economics model driven by digital transformation, says Bernard Golden, Capital One’s director of cloud strategy.
“Cloud is the new factory,” says Golden, who oversees cloud infrastructure for a company that runs significant portions of its business on AWS. The cloud’s capacity to automate functions and enable companies to operate more efficiently is analogous to Henry Ford’s revolutionization of the automotive industry via factory production, which automated several processes to enable faster production of cars, Golden says. Just as Ford must curtail the costs its machines incur in building its cars, enterprises must manage how they spend on cloud software.
Most enterprises today install cloud strategists and directors to help analyze and manage cloud resources. Figuring out how Nationwide should operate in the cloud is a big part of Joseph Daly’s role as director of cloud optimization.
As Nationwide moved more computing resources to AWS and Azure, it encountered challenges “rightsizing” cloud, says Daly, who cut his teeth in corporate tax accounting. This difficulty in selecting the right cloud service and instance for the tasks required is a common issue in moving to the cloud. So too is the cloud’s “black box,” Daly says, referring to the challenge of understanding costs for specific tech services. Cloud computing bills issued by AWS, Microsoft and others can be too detailed and unwieldy. FinOps helps enterprises track their spending on CPU, memory and storage, and then make adjustments, such as resizing servers to better align resources with requirements, says Daly.
Perhaps most importantly, FinOps helps Daly have a conversation with business peers about how much specific business processes cost to run in the cloud. His peers can then decide how they might change the process to reduce their cloud consumption bill. “That’s where you start to change the operating model,” Daly says. “It makes spending more transparent.”
Transparency in spending is critical for Alex Landis, finance manager and business partner for platform engineering for AWS at AutoDesk, where he is assisting in a major shift toward offering computer-assisted drawing software products via SaaS subscriptions on AWS and other clouds. A business analyst by training, Landis is learning how to run a footprint under the cloud.
For example, AutoDesk has put centralized management and governance policies in place to prevent engineers from, for example, blowing through $100,000 in fruitless cloud spending.
“I’m mostly concerned about internal usage of our cloud footprint, ensuring that we’re becoming more efficient, and doing the things that will make our customers more successful,” Landis says. That includes helping business executives understand the economics and value of cloud. Hammering out goals and business drivers is the key to successful cloud management, Landis says.
But the reality is that many people have made up their own FinOps practices on the fly. “It’s a wide-open new concept where best practices are being developed as we’re going,” Landis says.
Serverless poses new challenges
Linking dollars to services consumed in “cost utilization reports” produced by AWS has always been a struggle for IT and finance leaders, many of whom have found themselves “reverse engineering” their bills to match money spent to API calls triggered.
“The data ingestion is so great that you’re dealing in transaction data in hundreds of millions of rows,” says Jason Fuller, head of cloud management and operations of Here Technologies, which develops navigation software for automotive and other sectors.
But cloud vendors increasingly offer serverless and functions-as-a-service (FaaS), which allow companies to pay only for the code executed to run their applications. Those transactions often happen in sub-second response time, generating more line items more quickly, spurring a massive surge in billing data volumes and complexity. “When you write code for a new feature, the choices of services now are so awesomely fast and small, it allows you to spend so little money for a lot of people. But when you start running billions of things in 300 milliseconds, it quickly adds up,” Fuller says. “I think that is going to be the struggle.”
J.R. Storment, co-founder of Cloudability, which makes FinOps software that creates rate cards for enterprises consuming cloud services, concurs.
“We’ve seen monthly cloud billing files that get as large as 450 gigabytes uncompressed,” Storment tells CIO.com. “That’s a text-based cloud bill for a single company in a single month. Per-second billing coupled with SKU explosion [AWS offers more than 200,000 service SKUs] and broad adoption of public cloud into 9 figure a year territory at big enterprises has created a mass of data.”
To help codify FinOps practices, Cloudbility joined forces with Nationwide, AutoDesk, Here Technologies and others to launch the FinOps Foundation (F2), a non-profit trade association focused on promoting cloud financial management best practices and standards.
FinOps best practices
Regardless of whether firms adhere to the F2 guidelines or best practices of their own design, Capital One’s Golden says there are four key strategies for managing cloud spending:
- Determine who is spending on what. To do this, you need to “tag” your resources, or figure out what resources an application accesses. This is particularly important in enterprises in which thousands of people are accessing hundreds of applications.
- Don’t use a lawnmower when a hedge trimmer will do. Do you have the right EC2 instance for the computing workload you require, or are you overprovisioning? Only use and pay for what you need.
- Rationalize spend. Consider using reserved instances, a reservation of resources and capacity for a particular availability zone within a region. This enables you to buy computing horsepower you know you’ll need at a lower cost than on-demand procurement.
- Build apps suited for the cloud. If you’re a high-volume transaction business, that means designing your apps to scale horizontally, enabling you to add or subtract resources as needed.
“These are industry best practices around cost management,” Golden says. As for finding the right person to fulfill director of cloud strategy or cost optimization roles, Golden says it should be someone who is good at applying business requirements to what an enterprise needs to build. “You won’t find anyone with 10 years of experience managing cloud costs.”
This article originally appeared on CIO.com.