Cloud cost optimization is rapidly moving up the agenda as the dust settles on the technology decisions driven by the COVID-19 pandemic. Here are the key tools and principles to help developers stop cloud costs outpacing revenue.
To the concern of finance officers everywhere, the quick issuance of stay-at-home orders during the early days of the COVID-19 pandemic gave IT teams the authority to quickly scale up cloud usage to meet a rapidly changing set of employee and customer needs, almost overnight. In fact, cloud spending outpaced on-premises investments for the first time ever in 2020, marking a tipping point in enterprise consumption of cloud services.
Now, as the end of the pandemic hopefully comes into view, the party might be coming to an end, and cloud cost optimization will be on the agenda for many organizations as purse strings tighten.
“At the start of the pandemic, people were focused on getting stuff working. The cost of cloud was on the back burner,” 451 Research analyst Owen Rogers told InfoWorld. “Over the past three months there has been increased interest in managing that cost and optimizing for the future. The thing I am less sure about is when the dust will fully settle for enterprises.”
Where that cloud-cost level finally rests will vary from organization to organization, but according to Eugene Khvostov, vice president for product and engineering at cost-optimization specialist Apptio, “The core pains are the same: Who is responsible for what cost, how do I make them aware and accountable, and how do I scale that across the organization.”
What is for certain is that cloud costs can’t continue to grow out of line with revenue. This is less a story about cutting cloud costs than it is about right-sizing for your environment. The great cloud correction is upon us.
Why finops may help manage cloud costs
“Over the years, we’ve heard the same stories over and over again. Engineering teams spend more than they need to on cloud, with little understanding of cost efficiency. Meanwhile, finance teams struggle to understand and keep up with what teams are spending. Then, to top it off, leadership doesn’t have enough input into company spending—and sometimes doesn’t even show a willingness to influence priorities,” JR Storment and Mike Fuller wrote in their 2020 book Cloud Finops.
“Finops” is a divisive term that is essentially a new model for technology cost governance, where organizations use a set of techniques and tools to better plan, budget, and forecast cloud spending requirements. As Storment and Fuller wrote:
In the simplest terms, finops brings financial accountability to the variable spend model of cloud. But that description merely hints at the outcome. The cultural change of running in cloud moves ownership of technology and financial decision-making out to the edges of the organization. It flips long-held, forward-looking capacity planning methodology on its head to become rate-optimization analysis for technology that’s already been used. And it forces IT, finance, and business professionals to work together in unfamiliar ways. It’s an acceptance that the old ways of managing infrastructure aren’t just ineffective in cloud; they are irrelevant.
As Storment and Fuller alluded, a new hub-and-spoke model for cloud cost management is emerging, where a central group is tasked with finding optimization opportunities and negotiating the best rates with vendors, while engineers and product owners must take ownership of their own cloud costs.
As you would expect, online grocery delivery company Ocado saw a huge spike in demand for its services during the pandemic, and cost savings weren’t the priority. Instead, “it was scaling to meet demand,” Alex Howard Whitaker, principal cloud engineer at Ocado Technology, told InfoWorld.
To help manage that increased usage, Ocado tasked its central platform team with identifying company-wide savings opportunities, such as using spot instances for its more stateless applications and shutting off environments on the weekends. The central observability team also now incorporates cost in the information it provides individual product owners, who are increasingly responsible for the cost of their applications.
Why finops may not really help contain cloud costs
“Finops is flexible and built into the processes of the business, but I don’t think enterprises are at that stage yet, where cloud is still consumed on an ad hoc basis,” said 451’s Rogers. “Investigating the use of reserved instances, cutting out waste, and bringing in optimization tooling are the interesting things, as not many organizations will have the guts to make a cultural change of this scale while there is still so much uncertainty.”
In a recent survey of 304 IT and business decision-makers, cloud-management provider CloudCheckr found 63% of respondents still see cost management as a key area of improvement when it comes to their public cloud usage, with only 31% reporting that they monitor and optimize public cloud costs effectively. Or, as cloud billing consultant Corey Quinn told InfoWorld, “everyone says they have a handle on [cloud costs], ‘kind of, but we aren’t doing it well.’”