Executive summary
A major AI development company implemented Kubernetes to help modernize DevOps practices and improve Amazon EC2 utilization. However, to gain visibility into shared infrastructure costs and identify product margins more accurately, the company required a seamless and cost-efficient Kubernetes and cloud cost monitoring solution. After vetting different technologies, the company chose and integrated IBM Kubecost to deliver the Kubernetes and cloud cost insights and optimization it required. With IBM Kubecost, the company’s engineering team identified $750,000 in yearly savings within its first four months using the platform (without impacting application performance). The quickly growing company is also now able to much more accurately define annual revenue targets, spending, and margins.
The challenge
Founded in 2017, this AI development company was born in the cloud. However, until recently the company’s development style was slowed by several inefficient DevOps practices— including shuttling code between servers, performing hot patches, and back-patching edits made directly in production into lower environments. In 2023, the company adopted Kubernetes to establish best practices such as CI/CD, IaC, and well-maintained development and test environments. Achieving better utilization of its rapidly growing Amazon EC2 instances was also a major goal of the move to Kubernetes.
The company’s engineering team also wanted to be able to prove out the projected cost-efficiency and value of its Kubernetes-driven improvements. As part of that goal, the team focused on how to identify the company’s cost of goods sold (COGS) based on shared infrastructure expenses. With those metrics, the company could better understand the total costs of operating its solutions for customers and then align its pricing and margins more accurately. Considering the criticality of these FinOps-facing metrics and the high potential cost of managing cost monitoring internally, the company realized the best path forward was an effective external solution that could meet its requirements.
The solution
The company’s engineering team vetted several potential solutions but found most to be so expensive that they offered no savings compared to hiring internally. Open-core IBM Kubecost (self-hosted with a support license) emerged as the most capable and financially prudent path forward—and with an especially strong price point given the savings the team knew it would be able to extract using the platform. Implementation was straightforward, and IBM Kubecost was deployed within a month.
The results
Single pane of glass for spending
With the platform in place, the company now easily views its Amazon EC2 spending across all of its accounts from a single pane of glass. The dev team can quickly identify overprovisioning and poorly sized workloads, and they can provide real-time metrics to DevOps engineers to help flag and troubleshoot any anomalies quickly.
The company can also easily identify untagged workloads and attribute them to specific products or development activities—bringing full accounting to infrastructure that was previously opaque. Whereas tracking down that untagged infrastructure by looking into individual accounts would once have been a cumbersome and time-consuming process, IBM Kubecost brought visibility into all those costs from its single pane of glass, delivering impressive value beyond just Kubernetes monitoring.
$750,000 annual savings
Within four months of implementing IBM Kubecost, the company identified and clawed back $750,000 in Kubernetes-related cloud costs. The company can now more accurately estimate yearly revenue targets and spending, enabling more effective margins.
Positioned for future efficiency
With IBM Kubecost, the company is now prepared to continue to grow and scale efficiently in the cloud: the engineering team cites that even adding 100 more IBM Kubecost-monitored clusters today would be a straightforward process.
The team is looking forward to utilizing IBM Kubecost’s efficiency metrics to further increase cluster optimization, with a goal of raising the current 30-40% efficiency level in most clusters into the high 60s.
Summary
According to the customer, IBM Kubecost has exceeded expectations in a very short time.
“Historically, FinOps would periodically ask our team, ‘Why is your bill so big? Can you do something about it?’” said the company’s vice president and head of engineering. “Now with IBM Kubecost, we take a proactive approach of constantly asking that question ourselves and managing those expenditures, honestly pretty ruthlessly. IBM Kubecost has been instrumental in helping us identify where we’re spending money—not just on Kubernetes but on EC2 instances across our multiple accounts, and all from one single pane of glass. IBM Kubecost is phenomenally easier to navigate than AWS cost reports, leading directly to the $750,000 in yearly savings that we were able to target just this past quarter. At this point, IBM Kubecost has already done more than I ever expected to help us identify and lock down costs.”
Learn more
IBM Kubecost provides real-time cost visibility and insights for thousands of enterprise teams using Kubernetes, such as Audi, Coinbase, GitLab, and Under Armour. Get cost transparency by project or business divisions and reduce your spend by over 30% in your first 30 days. Learn more about IBM Kubecost
![ApptioCareers-Employee[1]](data:image/svg+xml;base64,PHN2ZyB4bWxucz0iaHR0cDovL3d3dy53My5vcmcvMjAwMC9zdmciIHZpZXdCb3g9IjAgMCAyMjUgMjc1IiB3aWR0aD0iMjI1IiBoZWlnaHQ9IjI3NSIgZGF0YS11PSJodHRwcyUzQSUyRiUyRnd3dy5hcHB0aW8uY29tJTJGd3AtY29udGVudCUyRnVwbG9hZHMlMkZlbGVtZW50b3IlMkZ0aHVtYnMlMkZBcHB0aW9DYXJlZXJzLUVtcGxveWVlMS1xZzVuenB5MHdoaWJtbzZvdmJqNGM3MHh0aTUzdmo1OHM0endwajh0dHMucG5nIiBkYXRhLXc9IjIyNSIgZGF0YS1oPSIyNzUiIGRhdGEtYmlwPSIiPjwvc3ZnPg== "ApptioCareers-Employee[1]")
