On-premises pricing

You run Kubernetes in your own datacenter, on bare metal, or in a colo — there's no cloud bill for Kubeadapt to read prices from. Tell Kubeadapt what a vCPU-hour and a GiB-hour cost in your environment and every cost number in the product reflects your real spend instead of the fallback GCP rates.

This guide walks the calculation, where to enter the rates in the dashboard, and what to do with mixed hardware.

1. Confirm prerequisites

You need:

• A Kubernetes cluster registered in Kubeadapt
• Operator access to the cluster in Kubeadapt (viewers can read the Pricing tab but not change it)
• An internal cost-per-resource number for vCPU, RAM, persistent storage, and optionally GPU

The data path from the agent to Kubeadapt Cloud is the same as for cloud-hosted clusters — see How Kubeadapt works. The only difference for on-prem is where the prices come from.

2. Calculate your resource costs

The four numbers Kubeadapt needs:

• vCPU: dollars per vCPU-hour
• RAM: dollars per GiB-hour
• Persistent Volume: dollars per GiB-hour
• GPU (optional): dollars per GPU-hour

The depreciation calculation that matches most finance models:

Server cost: $3,000 (24 cores, 128 GiB RAM)
Depreciation: 36 months
Operating hours per month: 720

vCPU cost = $3,000 / 36 / 720 / 24  = $0.00579 per vCPU-hour
RAM cost  = $3,000 / 36 / 720 / 128 = $0.00108 per GiB-hour

If your finance team already has a chargeback rate sheet, use those numbers instead — Kubeadapt cares about what the rest of your org bills against, not what the depreciation spreadsheet derives.

For reference, when an override is left blank Kubeadapt falls back to GCP us-central1 rates: vCPU $0.031611, RAM $0.004237, GPU $0.95, Persistent Volume $0.00005479 — all per hour.

3. Enter the rates in the dashboard

Open the cluster in Kubeadapt and go to Clusters → [your cluster] → Preferences → Pricing. The Pricing tab has four fields that map one-to-one to the numbers from step 2:

Type the rates into the fields and Save. Empty fields fall back to the system defaults, so you can override only the resources you care about and let the rest stay on defaults.

If you'd rather derive the rates from a monthly bill or a depreciation schedule than compute them by hand, expand the Pricing Calculator below the form. Enter the monthly cost and the cluster's installed capacity, then Apply — the calculator fills the four fields with the computed rates. Review and Save to persist them.

4. Verify

Open Cost Explorer in the dashboard. Cluster cost for the current day should reflect the new vCPU and memory rates within a few minutes; daily aggregates rebuild within a day.

If the numbers look 5–10× too high or too low, re-check the unit. vCPU is per vCPU-hour, RAM is per GiB-hour, persistent volume is per GiB-hour. The most common mistake is pasting a per-month or per-day rate into a per-hour field.

To start over, Reset to Defaults on the same tab removes every override for the cluster and falls back to the system defaults.

5. (Optional) Mixed hardware

The Pricing tab applies one rate set to the whole cluster. If you run multiple server generations on separate clusters, set different rates per cluster — each cluster has its own Pricing tab and its own overrides.

For tiered pricing inside a single cluster (e.g., premium and standard node pools sharing one cluster with different rates), contact support.

Next steps

• How Kubeadapt works — the data path from agent to dashboard, identical for on-prem and cloud
• Configuration overview — the full chart reference
• Right-sizing — once prices are accurate, the optimization recommendations are too