# Introducing the ScyllaDB Cloud Metering API: pull your own usage data
**URL:** https://forum.scylladb.com/t/introducing-the-scylladb-cloud-metering-api-pull-your-own-usage-data/5481
**Category:** Announcements
**Tags:** cloud, cloud-release
**Created:** [September 1, 2026, 8:27pm UTC](https://forum.scylladb.com/t/introducing-the-scylladb-cloud-metering-api-pull-your-own-usage-data/5481 "2026-09-01T20:27:08Z")
**Posts on this page:** 1
**Page:** 1
### Author:  [@Michael.Hollander](https://forum.scylladb.com/u/Michael.Hollander)
#### Post date: [September 1, 2026, 8:27pm UTC](https://forum.scylladb.com/t/introducing-the-scylladb-cloud-metering-api-pull-your-own-usage-data/5481/1 "2026-09-01T20:27:08Z")
We are happy to announce that we have shipped the **Metering API** for ScyllaDB Cloud!
You can now pull your own usage data - compute, network and backup - directly from an API, at node level.
It is generally available today for every ScyllaDB Cloud account.
## What it gives you
Usage quantities for the resources in your account:
- **Compute** - how long each node ran, per node, in node-seconds
- **Network** - traffic attributed to each network group
- **Backup** - backup storage and requests per cluster
Each resource comes back with a total for the period you asked for, plus a time series broken down by hour or day.
A note on scope: this reports **how much you used, not what you were charged**. Usage figures and invoices are produced by different systems and will not always line up exactly - if you have a billing question, contact support as you normally would.
## Making a request
`POST https://api.cloud.scylladb.com/billing/v1/metering`
Authenticate with a bearer token created on an account with the **ADMIN** role. The response always covers the account the token belongs to.
The complete reference - every field, enum value and error response - is in the API documentation: [ScyllaDB Cloud API → Metering](https://cloud.docs.scylladb.com/stable/api.html#tag/Metering). What follows here is the short version to get you started.
```bash
curl -X POST "https://api.cloud.scylladb.com/billing/v1/metering" \
-H "Authorization: Bearer " \
-H "Content-Type: application/json" \
-d '{
"from": "2026-08-01T00:00:00Z",
"to": "2026-09-01T00:00:00Z",
"granularity": "GRANULARITY_DAILY"
}'
```
| Field | Required | Notes |
| --- | --- | --- |
| `from` | yes | RFC3339. Cannot be in the future. |
| `to` | yes | RFC3339. Cannot be in the future. |
| `granularity` | no | `GRANULARITY_DAILY` (default) or `GRANULARITY_HOUR` |
| `clusterId` | no | Narrow to a single cluster |
| `dcId` | no | Narrow to a single datacenter |
| `networkGroupId` | no | Narrow to a single network group |
## What comes back
JSON, nested by cluster:
```auto
clusters[]
├─ clusterId, resourceName, deploymentTarget
├─ dcs[]
│ ├─ dcId, regionName
│ └─ nodes[] ← compute
├─ networkGroups[] ← network
└─ backup[] ← backup
every resource carries:
dimensions{} metadata (region, instance type, ...)
meteringQuantity total for the period
series[] { t, v } per bucket
```
## CSV export
Add `?format=csv` to the URL, or send `Accept: text/csv`, and the same request comes back as a flat table - one row per resource per time bucket, with a `Type` column separating compute, network and backup. It downloads straight into Excel or Sheets.
Two things to know about the CSV: compute quantities are converted from seconds to **hours** , while network and backup keep whatever unit the provider reported - so read the `Unit` column rather than assuming. And errors are always returned as JSON, never as CSV.
## Worth knowing
- **Network and backup data starts 1 May 2026.** Requests covering earlier dates return nothing for those two resource families. That is the start of the data, not a drop in usage - expect them to appear from May onward when charting a longer trend.
- **`dimensions` is deliberately free-form.** The keys differ between compute, network and backup, and between deployment types. Read it defensively; do not hard-code a fixed key list, as the set can change.
- **`networkGroups` entries are scoped to the group, not the cluster.** A network group can span several clusters. The entry appears under each attached cluster carrying the _whole group’s_ usage, so summing across clusters double-counts. Use `networkGroupId` to isolate one.
- **BYOA clusters return compute only.** Network and backup are not reported for Bring Your Own Account deployments.