How Armis eliminated manual resource tuning and improved workload stability at scale.
About Armis
Armis, now part of ServiceNow, is the asset intelligence cybersecurity company protecting the world’s most critical organizations. The Armis Centrix platform provides complete visibility and protection across IT, OT, IoT, and medical devices, securing Fortune 100, 200, and 500 companies, national governments, and critical infrastructure operators. When security is your product, your infrastructure has to match the reliability you sell.
The Challenge
The problem wasn’t what was running. It was how efficiently it was running.
At deployment time, Armis’s infrastructure team had no reliable way to know the right resource allocation for each workload. The team configured requests once and never revisited them. Those static estimates became the permanent production baseline regardless of how workload behavior shifted over time.
The consequences compounded: workloads running on guesswork from months ago, OOMKills surfacing when allocations ran short, scaling events requiring manual intervention. And as the environment grew, hand-tuning resource configurations became a workflow the team could no longer sustain.
The Solution
Armis deployed ScaleOps across all their clusters. Within hours of enabling optimization, ScaleOps had already started learning actual workload behavior and acting on it.
Pod rightsizing replaced deployment-time estimates with continuous, demand-driven allocation. The platform’s predictive, context-aware engine analyzed each workload independently, right-sizing resources based on what workloads actually needed rather than what the team approximated at deployment.
Manual tuning essentially disappeared from the workflow. ScaleOps ran continuously across the full cluster footprint. The infrastructure team monitored it rather than managed it.
The Results
Across all clusters, Armis reduced vCPU allocation by 30%. Workloads became more stable, OOMKills and scaling-related events dropped, and the team stopped spending time on resource configs that were out of date the moment they were written. With the platform handling it autonomously, developers got back to building the product.
“Manually tuning has essentially disappeared from our workflow. Developers can now focus on what matters most: building the product. Now it’s all handled autonomously.” — Lior Ran, Staff Infrastructure Engineer, Armis (now ServiceNow)