Cutting Cloud Costs

Visibility is the key to maintaining control.

The switch to remote work during 2020 saw a spike in demand for cloud-based services, for which companies could still be paying a premium.

According to data from market-intelligence firm Synergy Research Group, the spending on global cloud infrastructure services reached nearly $130 billion in 2020, way over the $89 billion spent on data center hardware and software, representing a 35% annual increase for cloud and a 6% drop for owned data centers.

Working from home fueled the sudden surge, but businesses have been migrating to the cloud—lured by promised agility and reduced maintenance and repair costs—for the past decade.

The cost of IT hardware previously caused companies to put the brakes on spending, but the cloud’s nebulous pricing has led to runaway costs for many businesses. Respondents to the Flexera 2021 State of the Cloud Report estimate their organizations waste 30% of cloud investments, while Flexera found that actual waste is 35% or more on average.

Cloud hosting typically is the highest or second-highest cost when running services or software, notes John Bonney, CFO at software-delivery platform provider Harness.

“Engineers all have access to cloud platforms, and they can purchase cloud resources without much governance,” he says. “There’s no purchase order and no real controls on that. So it’s like giving somebody the keys to the kingdom, but they don’t have to pay for it directly.”

It’s often difficult to keep tabs on who is iterating, upgrading or migrating which systems onto which cloud, agrees Mukund Rao, chief business officer of Banking, Financial Services and Insurance at technology consultancy Mindtree. “Once you have a process and system in place that gives you that view through a single pane of glass, and a mechanism to request more resources, overall spend trends are much simpler to forecast and control.”

Improving Cloud Coverage

Visibility is just one-third of the battle to control cloud spending, according to Rao. Tracking resource use and maintaining flexible spending are the rest of the fight. Ideally, resource tracking would be automated, or would at least red-flag issues when they occur; while the cloud-hosting providers would offer tools to optimize cloud resource consumption dynamically.

Reining in spending on cloud services does not have to be the responsibility of the finance team alone; it can be engineering’s responsibility as well, according to Harness’ Bonney.

“So, you’ve got accountability and ownership on the business side, which I’ll call engineering; and then you’ve got finance on the governance side, and they can work together,” he says. “Finance can see an alert in a materiality threshold of dollars; but then somebody’s got to act on that alert and do it quickly, and finance can’t do that.”

In mid-June, Harness released its Cloud AutoStopping feature for its Cloud Cost Management platform, enabling it to automatically stop resources the software deems unnecessary.

“We can automatically shut it down and reduce costs,” explains Bonney. “So, it’s not just transparency and immediacy—you also need automation to help manage cloud costs.”

Although the largest cloud providers have cloud management tools—AWS Cost Explorer for Amazon Web Services, GCP Billing for Google Cloud Platform and Azure Cost Management and Billing for Microsoft Azure—they provide only a window into their respective cloud services.

Meanwhile, on March 1, IBM introduced its IBM Cloud Satellite offering, which monitors the utilization of cloud resources across IBM’s and other providers’ cloud environments, via the platform’s abstraction capabilities.

“It provides a single interface for chief information officers to view consumption across cloud providers, to therefore ensure that they’re making the right decisions and their costs aren’t racking up,” explains Howard Boville, senior vice president, IBM Hybrid Cloud.

The Hybrid Conundrum

As businesses move more and more toward hybrid and multicloud environments for greater innovation and efficiency, controlling cloud costs can be even more challenging.

“The nature of large enterprises—projects, departments and service lines that each are on their independent transformation journeys—means that hybrid multicloud is a reality,” states Mindtree’s Rao. “It is inevitable—and that means resource waste is inevitable.”

The single largest challenge for CIOs is determining how to have visibility into the cloud environment to achieve cost control and manage unwieldy patchworks of IT real estate, all with their own objectives, budgets, timelines and benchmarks, he notes.

“It comes down to line of sight, both from a holistic perspective and at the granular level of each project,” Rao adds. “Do you have the right environment in the right place at the right time, deployed at the right speed, capacity and cost? The challenge and solution are in defining an architecture where you can manage these islands of clouds. Once an organization has a mechanism to at least see, if not operate, all this as one integrated system, you can start to see significant ROI [returns on investment].”

A case in point is an unnamed high-growth Harness client that witnessed its cloud costs rise from $100,000 to almost $2 million per year.

“Their massive growth hid it; but once we plugged in our cloud-cost platform, we found out that 70 to 80% of their cloud spend was unallocated,” says Harness’ Bonney. “Cloud had been spun up, used briefly and never shut down. They saved more than $1.5 million just dropping that back down to what was needed.”

For many companies, settling on just one cloud provider resulted in bad implementations, as the vendor’s only concern is moving clients to the cloud rather than understanding their business processes, according to IBM’s Boville.

“A pure-play cloud service provider has only one hammer in the toolbox; and therefore, everything looks like a nail,” he says. “We’ve got a bunch more tools and can take a far more artisanal approach in terms of how we craft solutions for you.”

As companies grow, the pressure clouds put on their margins “can start to outweigh the benefits,” conclude the authors of The Cost of Cloud, a Trillion Dollar Paradox, an article published by Silicon Valley venture capital firm Andreessen Horowitz.

As a  prime example, the authors cite Dropbox, which saved $75 million from 2016 to 2017 after the file-hosting service went to a hybrid environment and repatriated some workloads away from public cloud.

“The exact savings obviously varies [by] company, but several experts we spoke to converged on this ‘formula’: Repatriation results in one-third to one-half the cost of running equivalent workloads in the cloud,” they wrote. “Furthermore, a director of engineering at a large consumer internet company found that public cloud list prices can be 10 to 12 times the cost of running one’s own data centers.”

According to Bonney, not only gross margins or costs can be saved through cloud management; people can also save time, which companies can repurpose. “You can redirect their energy toward building new products and toward your customers,” he says. “That’s an ROI that’s hard to define, but it really is the icing on the cake—if you can reduce your cloud bill by 50%, and you may not need to hire another 50 engineers next year because you’ve just repurposed the existing ones.