Resource Usage Tracking

Ververica Platform tracks Compute Unit (CU) usage per namespace over time and exposes the data as a CSV report. This feature is informational only. It does not affect licensing, billing, autoscaling, or resource limits.

Overview

• Metric: CUs per namespace (fractional values supported, for example 1.2)
• Sampling frequency: every 1 minute
• Timezone: UTC only
• Records are written only when CU usage changes, not on every sample
• If CU usage is unchanged for 24 hours, the value is re-emitted as a new record

Data Model

Each record represents a time interval during which CU usage was constant.

Storage Behavior

Ververica Platform uses an interval-based model to store CU usage:

• When CU usage changes, a new record is written with the updated value.
• When CU usage stays the same for 24 hours, a new record is written with the same used value. This is the 24-hour re-emission rule.
• The 24-hour rule applies independently per namespace. One namespace re-emitting does not affect others.

Examples

Example 1: CU Changes Within Minutes

Namespace: payments

Stored records:

1begin=10:00, end=10:05, namespace=payments, used=2
2begin=10:05, end=10:20, namespace=payments, used=4
3begin=10:20, end=...,   namespace=payments, used=3

Example 2: CU Changes Across Hours

Namespace: analytics

Stored records:

1begin=08:00, end=14:00, namespace=analytics, used=10
2begin=14:00, end=18:00, namespace=analytics, used=20
3begin=18:00, end=...,   namespace=analytics, used=5

Example 3: Unchanged for Multiple Days

Namespace: abc, constant at 5 CU for 5 days.

Because of the 24-hour re-emission rule, one record is written per day even though the value does not change.

Stored records:

1Day1 09:12 → Day2 09:12 : 5 CU
2Day2 09:12 → Day3 09:12 : 5 CU
3Day3 09:12 → Day4 09:12 : 5 CU
4Day4 09:12 → Day5 09:12 : 5 CU
5Day5 09:12 → ...

Example 4: Multiple Namespaces

Namespaces a and b are tracked independently.

Namespace a changes at 15:00 so a new record is written. Namespace b is unchanged so it re-emits after 24 hours. The two namespaces do not affect each other.

Export Resource Usage

1GET /api/v1/status/resourceusage?from=YYYY-MM-DD&to=YYYY-MM-DD

• from is inclusive at 00:00 UTC on the given date.
• to is exclusive at 00:00 UTC on the given date.
• The response is a CSV file sorted by begin, then by namespace.

Example response:

1# Start: 2026-01-01
2# End:   2026-01-05
3begin,end,namespace,used
42026-01-01T09:12Z,2026-01-02T09:12Z,abc,5
52026-01-02T09:12Z,2026-01-03T09:12Z,abc,5
62026-01-03T09:12Z,2026-01-04T09:12Z,abc,5
72026-01-01T08:00Z,2026-01-01T14:00Z,analytics,10
82026-01-01T14:00Z,2026-01-01T18:00Z,analytics,20

Analyze Usage

To work with the exported CSV:

• Filter by namespace: select rows where the namespace column matches the namespace you want to analyze.
• Calculate CU-hours: for each row, multiply the used value by the length of the interval (end minus begin), then sum across all rows.
• Find peak usage: take the maximum used value from the filtered rows.

Delete Usage Data

This endpoint is available to admins only.

1DELETE /api/v1/status/resourceusage?to=YYYY-MM-DD

This deletes all records where begin is before the given date. By default, records newer than 365 days are protected and the request fails if any matching records fall within this window. Add &force=true to override the protection and delete records within the retention window.

Configuration

1vvp:
2  resourceUsageTracking: true

Set resourceUsageTracking to false to disable collection. No new data is collected while the feature is disabled. Existing data is not affected.