How Databahn Cut AWS Infrastructure Costs by 75% with DevZero
on multiple AWS clusters through intelligent node automation
consolidated 29–36 manual pools into 1–2 automated pools per cluster
dropping from 1,000's cores to 500-600
Kubernetes upgrades with automatic node reconciliation
About Databahn
Databahn is an AI-powered data pipeline management platform that helps global enterprises intelligently manage their telemetry data across security, IT, and observability systems. Founded in 2023 and headquartered in Plano, Texas, Databahn recently raised $17.2 million in Series A funding and has quickly emerged as a leader in data pipeline platforms.
The Challenge: AWS Infrastructure at Breaking Point
As Databahn scaled its multi-cloud infrastructure to support rapid customer growth, the engineering team faced mounting AWS challenges that threatened both business margins and team sustainability.
Explosive AWS Costs
Databahn's AWS clusters was running at an unsustainable $68 per hour, with costs 2.5 times higher than their comparable Azure environment, despite similar workloads. Cross-availability-zone placement of Kafka and Kubernetes nodes was generating massive EC2-other data transfer costs. With annual infrastructure spend increasing, leadership demanded an immediate 50% reduction in infrastructure costs with predictability. Databahn's infrastructure was severely overprovisioned.
Technical Complexity and Fragmentation
The infrastructure team was managing 29-36 dedicated EKS node pools with hard-coded node affinities, taints, and tolerations. Service alone was fragmented across 58 different nodes with 64 pods consuming 540 cores. The cluster autoscaler performed poorly at removing nodes, leading to persistent overprovisioning.
Services were experiencing lag spikes exceeding 1 million messages. Manual interventions became routine, with engineers working reapplying node taints and tolerations after infrastructure changes such as Kafka migrations.
Operational Paralysis
The team had no granular cost attribution to understand which workloads were driving AWS spend. Fear of disrupting production led to small, incremental changes, while leadership demanded aggressive cost cuts.
The Solution: DevZero's Kubernetes Automation for AWS
After evaluating both open-source and commercial alternatives, Databahn chose DevZero for its unique combination of speed, multi-cloud support, and Karpenter-inspired architecture, which resonated with its technically sophisticated team.
Rapid Implementation with Immediate Value
DevZero's installation took just 45 seconds with a single kubectl command. After 48 hours of data collection, the team was ready to begin optimization. Within 10 hours of enabling DevZero's policies, Databahn saw measurable cost reductions on its AWS infrastructure.
“I'm highly proficient in Kubernetes. I was a Kubernetes contributor back in 2020 and part of the Karpenter working group. We thought about going with open-source Karpenter, but I wasn't comfortable with the production outage risk. DevZero's Karpenter-based approach felt familiar, but with the safety and multi-cloud support we needed.”
Intelligent Node Automation for AWS
DevZero's node automation consolidated Databahn's sprawling manual EKS infrastructure. The 29-36 dedicated node pools with hard-coded configurations were replaced by 1-2 DevZero-managed pools per cluster. The system automatically selected optimal instance types, including ARM-based instances (C7G, M7G), mixed spot and on-demand capacity based on workload requirements, and consolidated nodes into a single availability zone to eliminate costly cross-AZ data transfer. In addition, the Databahn team had already invested in purchasing savings plans that they wanted to utilize - for this, DevZero’s node automation system was configured to respect and fully utilize current savings plans commitments first, before introducing on-demand or spot instances.
Understanding AWS Autoscaling Evolution
The evolution of Kubernetes autoscaling on AWS provides important context for DevZero's approach. Karpenter began as an open-source project, with AWS actively supporting it with dedicated engineering resources. As the project matured, AWS introduced EKS Auto Mode, a managed Karpenter solution that simplifies the operational overhead of running Karpenter directly.
While both open-source Karpenter and AWS EKS Auto Mode offer node provisioning automation, they ultimately represent revenue opportunities for cloud providers rather than cost optimization tools for customers. DevZero took a different approach, building on Karpenter's solid foundation while adding critical capabilities that the open-source and cloud-native tools lack.
Workload-Level Optimization
DevZero's vertical pod autoscaling (VPA) worked alongside Databahn's existing KEDA horizontal autoscaling setup on AWS. By taking ownership of resource requests and limits in ArgoCD deployments, DevZero automatically rightsizes workloads without disrupting the team's GitOps workflow. This fixed performance issues, such as spikes in Preview service lag, while dramatically reducing resource waste.
Cost Visibility and Attribution
DevZero provided the AWS cost visibility that Databahn desperately needed. Tag-based cost attribution enabled customer chargeback models, giving leadership the confidence to commit to AWS reservation plans. The dashboard exposed hidden cost drivers, such as cross-AZ data transfer and suboptimal instance type selection, that had been silently consuming budget.
The Results: Dramatic Transformation on AWS
DevZero delivered cost reductions that exceeded Databahn's aggressive targets. On an AWS cluster, DevZero achieved approximately 75% cost reduction, bringing infrastructure spending in line with revenue targets.
“We were essentially able to reduce the cost of that cluster by about 75%. On AWS, DevZero demonstrated they could achieve significantly higher savings than we initially thought possible.”
Improved Resource Efficiency
EKS cluster utilization improved from under 20% to 31% and continues climbing toward the target of 70-75%. Core provisioning was reduced by 40-50%, dropping from 1,000s cores to 500-600. Databahn's service, previously fragmented across 100s of nodes, was consolidated, delivering significant efficiency gains per cluster, per region.
Eliminated Operational Toil
The impact on team productivity was transformative. Kubernetes upgrades that previously required manual, careful node pool management now occur automatically with zero downtime.
DevZero eliminated these incidents entirely. Cluster upgrades that previously required careful manual node pool management became seamless, with node automation continuously reconciling and cycling nodes in the background.
The team saved an estimated 20-40+ hours per week by eliminating manual node scaling, after-hours firefighting, and emergency interventions. Engineers could finally focus on product development rather than on AWS infrastructure crisis management.
Strategic Business Impact
DevZero's cost visibility gave Databahn's leadership the confidence to negotiate long-term commitments with AWS. The ability to attribute costs to specific customers enabled the development of new pricing models. Most importantly, Databahn could now confidently present its infrastructure cost improvements as demonstrating clear progress toward the required margin targets.
Technical Implementation Details
DevZero provided a seamless integration process with AWS infrastructure:
- ArgoCD Integration: DevZero seamlessly integrated with Databahn's GitOps workflow. By taking ownership of the requests, limits, and replica count fields in deployments, DevZero can automatically optimize workloads while ArgoCD continues to manage the rest of the deployment configuration.
- KEDA Coexistence: DevZero's VPA works alongside KEDA's HPA. KEDA continues to handle horizontal scaling based on Kafka lag (with a 5M-message threshold), while DevZero optimizes vertical resource allocation for each pod.
- Terraform Modules: Databahn standardized their AWS infrastructure provisioning with Terraform modules that incorporate DevZero's tagging requirements from day one, ensuring all new EKS data planes have proper cost attribution built in.
- AWS Cost Explorer: DevZero's tagging strategy enabled detailed cost analysis in AWS Cost Explorer, allowing the team to track infrastructure costs at the customer and workload level.
Multi-Cloud Architecture
While this case study focuses on Databahn's AWS implementation, DevZero's multi-cloud capabilities were crucial to the decision. Databahn operates:
- 10+ data planes on AWS
- 5+ data planes on Azure
- Considering deploying dataplanes on GCP and OCI
Consistent Automation Approach
The consistent automation approach across all cloud providers means Databahn's team doesn't need to learn different tools or approaches for each cloud. This was particularly important given their plans to expand to GCP and consolidate multiple data planes in the future.