SaaS agent bundle

Lattice Kubernetes Meter

Allocates cluster costs to teams using namespace budgets, idle node detection, and daemonset fairness rules you can defend in meetings.

₩6,100,000 per year (informational)

Rolling annual with monthly health checks

Overview

Lattice listens to kube-state metrics and cloud billing joins to highlight silent CPU reservations, orphaned PVCs, and autoscaler thrash. It is opinionated about label contracts, but every opinion links to a playbook snippet. Designed for midmarket shops that outgrew spreadsheets but refuse a six-month RFP for megavendor APM add-ons.

Capabilities

  • Namespace burn rates with soft and hard thresholds
  • Spot interruption correlation for batch queues
  • GPU time-slicing summaries for shared pools
  • Chargeback stubs that map to general ledger codes
  • Helm release attribution with version drift notes
  • Daemonset amortization across tenant pods
  • Weekly digest tuned for platform guilds

Outcomes teams track

  • Cluster owners see idle headroom before finance escalates.
  • Tenant teams understand their share of control-plane overhead.
  • Autoscaling mistakes become visible within days, not invoices.

Plan questions

EKS, GKE, and AKS clusters running supported Kubernetes versions listed in the technical appendix.

Experience notes

  • Finally reconciled why our staging namespaces looked cheaper than production—turns out they were starving the daemonsets.