Problem
Unnecessary log groups, noisy log streams, and broad retention defaults were creating avoidable CloudWatch ingestion and storage charges.
Production cost saves across AWS, EKS, CloudWatch, VMware, Oracle, and Kubernetes platforms
These stories show how I approach infrastructure cost optimization: trace the real usage pattern, protect reliability, reduce waste, and turn the work into measurable monthly savings.
AWS FinOps
Kubernetes Platform Engineering
Remote SRE Leadership
CloudWatch Logs Cost Optimization
Reduced CloudWatch Logs Spend by $10K per Month
CloudWatch Logs costs were rising because applications and platform components were sending high-volume logs into groups that did not all carry equal operational value. I investigated the ingestion pattern, separated useful signals from avoidable noise, and reduced recurring spend while preserving the logs needed for debugging and incident response.
Open standalone case studyAction
Reviewed log group usage, removed unused or low-value ingestion paths, tightened retention, and aligned logging volume with real operational needs.
Impact
Saved approximately $10K per month on AWS CloudWatch bills without sacrificing critical observability for production systems.
- Reduced avoidable CloudWatch Logs ingestion
- Improved log group ownership and retention discipline
- Protected debugging and incident-response visibility
- Connected observability decisions to measurable business savings
EKS Data Transfer Cost Optimization
Cut Inter-AZ and Cross-Region Traffic Costs by up to $25K per Month
AWS EKS clusters were generating expensive inter-AZ and cross-region data transfer charges. I analyzed service communication paths, workload placement, and traffic behavior to identify where network cost was being created unnecessarily.
Open standalone case studyProblem
Kubernetes services and workloads were communicating across availability zones and regions more often than needed, increasing AWS transfer spend.
Action
Investigated EKS traffic flows, optimized workload placement and service communication patterns, and reduced unnecessary cross-zone and cross-region paths.
Impact
Reduced AWS data transfer costs by up to $25K per month while keeping reliability and production traffic requirements in view.
- Reduced inter-AZ and cross-region AWS data transfer charges
- Improved visibility into EKS network cost drivers
- Optimized Kubernetes service and workload communication
- Balanced cost savings with reliability and traffic locality
AWS Direct Connect Cost Optimization
Learned to Right-Size AWS Direct Connect Capacity for Lower Recurring Cost
AWS Direct Connect is easy to treat as fixed infrastructure, especially when it supports production traffic. I learned to approach it as a FinOps and reliability decision: first understand real port utilization and traffic growth, then validate the redundancy design before changing capacity. That made it possible to reduce recurring Direct Connect cost without making availability a trade-off.
Open standalone case studyProblem
Provisioned Direct Connect capacity and recurring port cost no longer clearly reflected observed traffic demand, but any change had to preserve critical connectivity and failure tolerance.
Action
Reviewed utilization trends, peak periods, traffic paths, and redundancy requirements. Used those findings to right-size capacity and remove cost that was not supporting a real operational need.
Impact
Reduced recurring AWS Direct Connect spend while retaining a resilience-first design and a clearer process for reviewing connectivity cost over time.
- Connected AWS billing data with port utilization and traffic trends
- Validated redundancy and failure scenarios before reducing capacity
- Right-sized Direct Connect for actual demand rather than historic assumptions
- Created a repeatable FinOps review for ongoing connectivity cost control
VMware Oracle Licensing Optimization
Reduced Oracle Licensing Costs on VMware Clusters
Oracle workloads running across broad VMware clusters can increase licensing exposure when placement is not tightly controlled. I reduced that exposure by segregating ESXi hosts dedicated to Oracle VMs into a separate cluster and applying host affinity rules so Oracle workloads stayed on approved capacity.
Open standalone case studyProblem
Oracle VMs needed clearer infrastructure boundaries so licensing scope did not extend unnecessarily across general-purpose VMware ESXi hosts.
Action
Segregated Oracle-capable ESXi hosts into a dedicated VMware cluster and added host affinity rules to keep Oracle VMs on the intended hosts.
Impact
Reduced Oracle licensing cost exposure, improved audit readiness, and gave infrastructure teams clearer control over where Oracle workloads could run.
- Separated Oracle VMs from general-purpose VMware clusters
- Controlled placement with ESXi host affinity rules
- Reduced unnecessary licensing scope and cost exposure
- Improved governance for virtualized database workloads
More Production Impact
Reliability, migrations, and operational recovery
Alongside FinOps work, I have led incident recovery, cloud migration, observability, and platform automation work across remote-first teams.
Recovered Business-Critical MongoDB Systems
Restored MongoDB performance and application stability after critical indexes were deleted during a security incident.
Led DigitalOcean to AWS Platform Migration
Moved a production platform from DigitalOcean to AWS to improve scalability, operational maturity, and long-term infrastructure control.