Cloud’s power to digitally transform business comes with a major caveat: access to practically unlimited compute can be a budget buster.
Despite a broad enterprise push to optimize spending and commitments from AWS and other major providers to help, cloud customers passed on billions in potential savings last year, according to ProsperOps.
The FinOps solution provider analyzed $1.5 billion in AWS bills from hundreds of organizations over a 12-month period and found that more than half failed to take advantage of built-in discounts. Most left money on the table by paying on-demand rates for compute services rather than leveraging discount instruments, the report said.
Hyperscalers like AWS, Microsoft and Google Cloud offer a range of commitment-based discounts to help organizations manage their cloud costs, but using these savings instruments can be difficult, Chris Cochran, co-founder and CEO of ProsperOps, said.
Controlling cloud bills became the top cost concern for companies in 2023, frustrating finance executives as spending on tech services and software consumed nearly half of global IT spending, a number Gartner expects to surpass $5 trillion this year.
“During the early phases of cloud computing, the common refrain was ‘security is everyone’s responsibility,’” Drew Firment, chief cloud strategist at Pluralsight, said. “As cloud bills continue to rise, organizations are now singing from the top of their headquarters that ‘cost is everyone’s responsibility.’”
While cyber remains a perennial migration hurdle, more organizations are turning to FinOps practices to rein in runaway costs. But implementation is a work in progress.
“If you poll the average cloud user, they’ll say that security and cost are the top two things they worry about,” Cochran said in an interview with CIO Dive. “But when it comes to cost, 30 cents on every dollar is wasted. That’s been true for the last decade and the number hasn’t gone down.”
Effective savings rate
Usage-based pricing and scalability can obscure the true cost of cloud, particularly for those relatively new to the technology. Billing complexity is a compounding factor.
“A lot of the clients we talk to are still dealing with cloud 101 and they're being asked to do some pretty advanced maneuvering,” Tracy Woo, Forrester principal analyst, said. “It's not only understanding instances and services and standardizing and normalizing them, but it’s also understanding all the different boundaries and restrictions that affect price.”
ProsperOps focused its initial product solution on AWS, the largest of the global hyperscalers, and singled out compute rather than data for its analysis. The company used the effective savings rate metric which compares actual spending with discounts to the standard rate for equivalent compute workloads.
Cost metric rationalization is spreading. AWS embedded the FinOps metric in the billing and cost management dashboard it rolled out in November as part of a broader optimization push.
“The level of detail AWS provides in our billing and usage analysis equips our customers with more opportunities to save, optimize and increase efficiency,” Chris Hennesey, AWS enterprise finance strategist, said in an email. “However, we find that customers are most successful with these efforts together with having right sized their environment — the process of matching instance types and sizes to workload performance and capacity requirements.”
In November, the FinOps Foundation expanded its cloud billing standards specifications, called FOCUS, shortly after AWS joined Microsoft, Google Cloud and Oracle on the project’s steering committee.
The move reflected a broader shift in the dynamic between users and providers, Woo said. “These initiatives are the end users saying to the cloud providers: ‘We’re exhausted. You fix this for us. We understand this is complex, but you are the ones who should be standardizing.’”
Leveraging cloud deals
Billing specifications can help end users sort through past spending, a big step toward optimization. Finding areas to leverage discounts and identifying workloads that can be run for less are logical next moves.
ProsperOps found that roughly half of cloud spend during the period analyzed went into compute, a category teeming with poorly planned provisioning and rife with optimization opportunities.
“We focused on AWS because it’s the most complete cloud and everybody uses it,” Cochran said, comparing the hyperscaler’s array of enterprise offerings to Lego products.
“AWS throws all their best Legos into a big bucket and you can use those Legos in very powerful ways, assuming you can put them together and apply them in your environment to get the outcomes you need,” said Cochran.
AWS gives cloud financial management teams three main savings levers, the report said: commitment-based discounts, volume-based discounts and enterprise discount programs.
Only 38% of organizations analyzed by ProsperOps took advantage of the one- to three-year savings plans AWS offers, and that was the most commonly used discount instrument. Standard Reserved Instances were used by 18% of AWS customers and Convertible Reserved Instances by just 14%.
AWS is actively talking to customers about savings plans and reserved instances and encouraging them to use the compute optimizer dashboard, Hennesey said.
Enhanced metrics, built-in dashboards and clearer billing terms have made provisioning easier, according to Brian Adler, senior director of cloud market strategy at Flexera.
“When reserved instances first came out, the joke we used to make is to use a reserved instance you need to have a PhD in economics and a Ouiji board to understand complex mathematical financial modeling and be able to predict the future,” Adler said. “You don't need to do that with savings plans anymore, yet, folks are still not using them.”
