Stefanini Cuts Kubernetes Compute Costs 26% with PerfectScale by DoiT

The design choice left the platform team accountable for the cost, efficiency, and reliability of everything running underneath.

The existing toolset fell short of providing the visibility and granularity the team needed to understand what was happening inside the clusters. The solutions they evaluated weren't enough to identify meaningful optimization opportunities or provide the level of insight required to manage their Kubernetes environment effectively. A monthly cloud bill shows compute expense, but not which workloads drive it, how efficient they are, or where the waste sits. Engineers reused resource requests and limits from previous deployments rather than sizing each workload, so clusters filled with over-provisioned pods. The same blind spots created reliability risks: workloads that were CPU-throttled or restarting because their resources were never set correctly. As AI adoption pushed cloud consumption higher every month, that lack of visibility became a growing financial and operational problem.

Stefanini evaluated both open source and commercial options. PerfectScale by DoiT stood out for exposing what was happening inside the clusters in the clearest, most actionable way, and for doing it in a form the platform team and individual engineering teams could both use. That let Stefanini democratize optimization data across the organization rather than keeping it locked with a handful of specialists.

The engagement began as a visibility play in September 2023. PerfectScale gave the team a workload-level view of cost and waste over time, including the peaks, so they could pinpoint exactly what to right-size. Automation was a secondary consideration at first, and it has since become the most valuable part of the tool. PerfectScale now continuously right-sizes workloads automatically, and the team is extending optimization to the nodes themselves, choosing the best node pools and reserved-instance-aware VM types through a newer capability they are actively rolling out.

For David Adkins, the differentiator is the relationship behind the product. "PerfectScale was the best tool of all the ones we looked at," he said. "It provided that level of detail, and it did it in a way that was very easy to understand and take action on. The DoiT support team knows Stefanini's environment, proactively flags areas worth a closer look each month, and turns customer suggestions into shipped features, an active partnership rather than an invoice and a support queue.”

With AI enablement driving what Stefanini expects to be an explosion of new workloads on the platform, the team is leaning on PerfectScale to keep that growth efficient: capturing the remaining savings potential, expanding node-level optimization, and continuing a partnership David values for one simple reason. "When we suggest something important to us, it shows up in the product within a couple of months," he said. "It's not just a roadmap item."