How Databahn Cut Azure AKS Costs by 64% with DevZero
on Azure AKS infrastructure
Results visible within the first day of deployment
Consolidated complex manual infrastructure into 1–2 automated pools per cluster
Eliminated premium SSD waste on storage volumes
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 security data pipeline platforms.
The Challenge: Azure AKS Infrastructure Crisis
As Databahn scaled its multi-cloud infrastructure, its Azure AKS environment faced critical technical and financial constraints that threatened business operations.
Runaway Azure Costs
As the business scaled, Databahn's Azure clusters grew by 2-3x month over month. Additionally, the team discovered that premium SSDs were configured to run on their Kubernetes worker nodes, resulting in $1000s per week in unnecessary storage costs on top of 10s of $1,000s in weekly compute spend.
Pod CIDR Exhaustion
Databahn's Azure cluster reached a critical technical limit: they had exhausted their pod CIDR allocation, with 234 of 256 nodes allocated. The cluster was stuck on Kubernetes 1.30 because expanding the pod CIDR required a complex migration that the team couldn't risk in production. IP address starvation in their main subnet created additional deployment blockers.
DevZero helped identify a critical blocker during the initial assessment: Kubernetes version upgrades couldn't proceed until the pod CIDR problem was resolved.
Azure-Specific Complexity
Additionally, cross-regional traffic was generating massive data transfer costs. Deprecation of the storage class mid-upgrade caused backup service failures. The team had 29-36 dedicated node pools with hard-coded configurations, and bespoke Azure CLI commands were required for operations not available in the portal.
The Solution: DevZero's Multi-Cloud Kubernetes Automation
After evaluating alternatives, Databahn chose DevZero for its unique combination of rapid implementation, deep Azure expertise, and multi-cloud support, which would protect its investment as it expanded its Azure footprint alongwith the rest of the business.
Immediate Implementation Success
In the first phase, DevZero optimized workloads in a single cluster to reduce costs from $68 to $60 per hour (13-14% savings), and consolidated them into a smaller set of availability zones while still maintaining high availability. The initial savings came from optimizing just three to four services. Within 10 hours of implementing DevZero's policies, Databahn realized measurable cost reductions, demonstrating how quickly the platform delivered value with minimal infrastructure changes.
Understanding Azure's Autoscaling Approach
The Azure Kubernetes ecosystem has evolved in a manner similar to AWS. Karpenter, originally an open-source project, was supported by Azure with azure-karpenter for a period. As with AWS, Azure eventually introduced Node Autoprovision (NAP), their managed autoscaling solution designed to reduce operational complexity.
However, both open-source Karpenter and Azure's NAP are ultimately designed to increase Azure consumption rather than optimize customer costs. DevZero takes a fundamentally different approach, leveraging Karpenter's foundation while adding capabilities specifically designed for cost optimization and operational simplicity.
DevZero's Azure-Optimized Intelligence
“The consistency across clouds was crucial for us. We're already on Azure and AWS, with GCP and OCI on the roadmap. Learning different autoscaling tools for each cloud wasn't sustainable. DevZero gave us one approach that works everywhere.”
- Better workload understanding: DevZero analyzes workload behavior patterns over time, not just current resource requests, enabling smarter scaling decisions that prevent overprovisioning.
- Simplified AKS cluster upgrades: DevZero's automatic node reconciliation and cycling resolved Databahn's pod CIDR constraints, enabling seamless Kubernetes version upgrades with zero downtime.
- Integrated workload and node optimization: DevZero's MPA works closely with node provisioning to ensure pods and nodes are right-sized together rather than treated as separate problems.
- Multi-cloud consistency: Unlike Azure-karpenter or NAP, DevZero provides the same automation approach across Azure, AWS, and other clouds, enabling teams to leverage expertise across their entire infrastructure.
Intelligent Node Automation for AKS
DevZero consolidated Databahn's 29-36 manual node pools into 1-2 DevZero-managed pools per cluster. The system automatically selected optimal VM types, mixed spot, on-demand and reserved capacity, and consolidated nodes into a smaller number of availability zones to eliminate cross-AZ data transfer costs that had been hidden in Azure billing.
“As we moved to DevZero, there are only one or two node pools, and that too is managed automatically. No more manual node pool creation, no more juggling dozens of pools with hard-coded taints and tolerations.”
Cost Visibility and Azure-Specific Optimization
DevZero's dashboards exposed Azure-specific cost issues the team hadn't known existed. Premium SSD defaults were immediately flagged. Cross-region data transfer costs became visible and actionable. Tag-based cost attribution enabled customer chargeback models, giving leadership confidence to commit to Microsoft's annual spend tiers.
The Results: Azure Infrastructure Transformed
DevZero achieved ~72% cost reduction on Databahn's Azure infrastructure, bringing spending in line with revenue and enabling the margins the team targeted.
Resolved Technical Constraints
DevZero's node automation solved the pod CIDR exhaustion problem that had blocked Kubernetes upgrades. By efficiently managing the node lifecycle and reducing the number of nodes required, Databahn could upgrade from Kubernetes 1.30 to newer versions, enabling access to expanded pod CIDR capabilities.
Operational Excellence Restored
Manual node management became automated. The team saved an estimated 20-40+ hours per week previously spent on infrastructure firefighting.
“DevZero gave us more than Azure cost savings. It enabled our engineers to focus on product innovation, provided the financial visibility and stability, and created a foundation for multi-cloud expansion that can scale with our growth. With DevZero managing our infrastructure complexity, we can focus on product development.”
Technical Implementation on Azure
DevZero provided a seamless integration process with Azure infrastructure:
- ArgoCD GitOps: DevZero integrates seamlessly with Azure-hosted ArgoCD deployments, taking ownership of resource optimization while preserving GitOps workflows.
- Azure CNI Overlay: DevZero's optimization strategies work within Azure CNI constraints, helping teams work around pod CIDR limitations.
- Storage Class Management: DevZero identified and helped migrate away from expensive premium SSD defaults, saving tens of thousands per week.
- Terraform Azure Modules: Databahn standardized AKS provisioning with Terraform modules that incorporated DevZero tagging from day one.