Skip to content

Volumetric Check Comparisons

The comparison decides what a Volumetric check measures and what Min and Max bound. There are three: Absolute Value bounds the row count itself, while Absolute Change and Percentage Change bound how much the count moved since one Measurement Period earlier.

Different from the Metric check comparisons

The Metric check uses the same three comparison names, but its behavior is entirely different. A Metric check measures a field value and compares it with the previous Scan. A Volumetric check measures the container's row count and compares it with the daily counts of earlier calendar days. Do not carry limits or expectations from one check to the other.

Absolute Value

Absolute Value asks: is the row count inside [Min, Max]?

  • What is compared. With a Measurement Period of 1 day, the row count the Scan reads. With a longer period, the average daily row count across the period: the count the Scan reads plus the daily counts of the previous N - 1 days, divided by N.
  • What Min and Max mean. Row counts.
  • Anomaly rule. A value below Min or above Max raises an anomaly. Both limits are inclusive.
  • Best for. Containers whose size has a known, stable range, such as a reference table or a snapshot that is fully reloaded each day.

Example. Min 100, Max 1,000, Measurement Period 1 day:

Scan Row count
Monday 640
Tuesday 45
Wednesday 980

Absolute Change

Absolute Change asks: did the row count move by more than expected?

  • What is compared. The row count the Scan reads against the daily count from exactly one Measurement Period earlier. A 1 day check compares against yesterday, a 7 day check against the same weekday last week. The days in between are not part of the comparison.
  • What Min and Max mean. Row movements. A negative Min allows the count to fall by up to that amount, and a positive Min requires growth of at least that amount.
  • Anomaly rule. If the count is below the earlier count plus Min, or above the earlier count plus Max, an anomaly is raised. Both limits are inclusive.
  • Best for. Containers that grow over time, where the size itself drifts but each load cycle adds a predictable amount. AI Managed Volumetric checks always use this comparison.

Example. Min 30, Max 70, Measurement Period 1 day:

Scan Earlier count Row count Change
Monday 100 150 50
Tuesday 150 220 70
Wednesday 220 300 80

Percentage Change

Percentage Change asks: did the row count move by more than a given percentage?

  • What is compared. The same pair as Absolute Change: the row count the Scan reads against the daily count from one Measurement Period earlier.
  • What Min and Max mean. Percentages of the earlier count. The form labels them Min Percentage and Max Percentage and takes values in percent (enter -20 for a 20% decrease). Through the API they are decimals (-0.2).
  • Anomaly rule. If the change, as a percentage of the earlier count, falls below Min or above Max, an anomaly is raised. Both limits are inclusive.
  • Best for. Containers whose size varies widely, where a fixed number of rows would be too strict for a large table and too loose for a small one.

Example. Min -20, Max 50, Measurement Period 1 day:

Scan Earlier count Row count Change
Monday 100 150 50%
Tuesday 150 120 -20%
Wednesday 120 65 -45.83%

Choosing a Comparison

The expectation you want to enforce Comparison
The container always holds between X and Y rows Absolute Value
The average daily size over a period stays between X and Y Absolute Value with a longer Measurement Period
Each load cycle adds at least, or at most, a number of rows Absolute Change
The size never jumps or drops by more than a percentage Percentage Change

The comparison is chosen when the check is created and is locked after the first Scan that includes the check. See How It Works.

See Also

  • Anomaly Reporting

    What the Shape Anomaly reports, how to read the numbers in the message, and why there is nothing to report per row.

    Anomaly Reporting

  • Examples

    Three production scenarios: a load that adds no rows, a weekly growth band, and a fixed volume range.

    Examples

  • Best Practices

    Guidelines for choosing the comparison and Measurement Period, scheduling the Scan, and working with AI Managed checks.

    Best Practices

  • Permissions

    The team permission each action needs: view, create, edit, archive, restore, and delete.

    Permissions

  • How It Works

    The complete reference: definition, when the check is evaluated, the properties, the history it needs, and AI Managed Volumetric checks.

    How It Works