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)
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.