Advanced Computing in the Age of AI | Saturday, January 28, 2023

Cloud Transition Begets Cloud Waste 

The more companies spend on cloud resources, the greater the likelihood those investments will be wasted on excess and therefore idle capacity along with the overprovisioning of infrastructure services, according to cloud optimization study.

The analysis by cloud resource management specialist ParkMyCloud estimates that enterprises will waste as much as $14.1 billion in cloud resources during 2019. Among the reasons is idle resources, underscored by the fact that public and other cloud resources are seldom used every day, around-the-clock, yet cloud pricing is largely based on that scenario.

“When you spin up a cloud resource in a public cloud provider such as Amazon Web Services, Microsoft Azure, Google Cloud Platform or Alibaba Cloud, it is always running unless you specifically turn it off,” ParkMyCloud notes on its web site.

Cloud waste is likely to grow as companies begin to move core business applications to the cloud. For example, public cloud spending is expected to top $206 billion in 2019, with infrastructure services growing at a 27.6-percent clip, according to estimates cited by ParkMyCloud, which is based among the sprawling data centers popping up in the northern Virginia suburbs of Washington, DC.

The company estimates that about two-thirds of cloud resources are devoted to computing, accounting for an estimated $26.3 billion in cloud spending. “This segment of [cloud spending] is especially vulnerable to waste, particularly from idle resources and oversized resources,” the company noted in a blog post.

Even as more cloud workloads move to production, including more frequent updates of applications, cloud computing resources are wasted in non-production but necessary steps like DevOps, testing and quality assurance. ParkMyCloud estimates that about 44 percent of cloud computing budget is devoted to these operations, providing little return on investment.

The other key contributor to cloud waste is “oversized infrastructure.” For example, industry studies have found that 40 percent of cloud instances are larger than required for their workloads. ParkMyCloud reported that infrastructure managed on its platform showed an average CPU utilization rate of 4.9 percent, indicating that even without accounting for memory resources, the infrastructure estimate above may be too conservative.

Either way, the varying estimates of underutilized capacity “paints a picture of gross underutilization, ripe for right-sizing and optimization,” the cloud management vendor asserted.

As more enterprises shift from operating their own servers to buying cloud services, market analysts such as Gartner Inc. (NYSE: IT)

predict cloud pricing will shift from hourly rates to “pay as you go” models. That transition also may fuel the rise of cloud resource management outfits like ParkMyCloud offering services designed to rein in cloud waste.

RightScale, another cloud resource management service, recently pegged cloud waste at $10 billion annually. Indeed, wasted cloud resources accounted for an estimated 35 percent of public cloud spending, RightScale revealed in its annual cloud report.

About the author: George Leopold

George Leopold has written about science and technology for more than 30 years, focusing on electronics and aerospace technology. He previously served as executive editor of Electronic Engineering Times. Leopold is the author of "Calculated Risk: The Supersonic Life and Times of Gus Grissom" (Purdue University Press, 2016).

Add a Comment