App · 2025
Kubestat
kubestat.dev · app I built · Founder and engineer
A reliability and cost console for EKS: which workloads are over-provisioned, which are one node failure from an outage, and what each namespace actually costs per week.
31%
average cluster spend removed
340
clusters observed
11
paying teams
The problem
After migrating production from ECS to EKS I could see cluster metrics and I could see the AWS bill, but nothing connected a specific Deployment to the money it burned or the risk it carried.
What I did
- Joined Prometheus workload metrics with Cost and Usage Report data to attribute spend down to a Deployment.
- Built a risk score from replica spread, PDB coverage, node-group diversity and request-versus-limit ratios.
- Added right-sizing recommendations with a one-click patch that opens a pull request against the manifest repo.
- Weekly digest email so teams see their own drift without opening the tool.
Architecture
Java and Spring Boot backend, React front end, Prometheus remote-read plus AWS CUR ingestion into Postgres and Timescale, agentless install as a read-only IRSA role, deployed by Helm into the customer's own cluster.
JavaReactPrometheusPostgresHelmEKS
Request path
Similar problem on your side?