Volumetric Check

Volumetric Check ensures data stability by monitoring dataset size fluctuations in rows or bytes. It detects anomalies by comparing current volumes against historical trends (daily, weekly, monthly). Users can configure rules for precise control, while automated threshold adjustments enhance accuracy over time.

Let's get started 🚀

Configure Volumetric Check

Step 1: Login into your Qualytics account and select the datastore from the left menu on which you want to add a volumetric check.

Step 2: Click the Add button and select Checks.

Step 3: A modal window appears. Enter the required details to configure the Volumetric Check.

Step 4: Enter the details to configure the volumetric check:

3. Comparison: Specifies the type of comparison: Absolute Change, Absolute Value, or Percentage Change:

4. Measurement Period Days: Enter the number of days for measurement.

5. Threshold: At least the Min or Max value must be specified, and including both is optional. These values determine the acceptable range or limit of change in the field's value.

Min Value

• Represents the minimum allowable increase in the field's value.

• A negative Min Value signifies an allowable decrease, determining the minimum value the field can drop to be considered valid.

Max Value

• Indicates the maximum allowable increase in the field’s value, setting an upper limit for the value's acceptable growth or change.

Step 4: After completing all the check details, click on the "Validate" button. This will perform a validation operation on the check without saving it. The validation allows you to verify that the logic and parameters defined for the check are correct. It ensures that the check will work as expected by running it against the data without committing any changes.

If the validation is successful, a green message will appear saying "Validation Successful".

Step 5: Once you have a successful validation, click the "Save" button.

After clicking on the “Save” button your check is successfully created and a success flash message will appear saying “Check successfully created”.

How It Works

The system automatically infers and maintains volumetric checks from the row count changes it observes over daily, weekly, and monthly windows. These checks enable proactive management of data volume trends, ensuring that any unexpected deviations are identified as anomalies for review.

Automating Adaptive Volumetric Checks

The following Volumetric Checks are automatically inferred for data assets with automated volume measurements enabled. All three use the Absolute Change comparison, so their Min and Max thresholds bound how far the row count may move across the check's Measurement Period rather than what the count itself may be:

• Daily (1 Day): the expected change in volume from one day to the next, inferred from the daily changes observed over the previous 7 to 28 days.

• Weekly (7 Days): the expected change in volume from one week to the next, inferred from the weekly changes observed over the previous four weeks.

• Monthly (28 Days): the expected change in volume over 4 weeks, inferred from the 4-week changes observed over the previous sixteen weeks.

In each case the thresholds are the range of changes actually observed during the training period, widened by a multiple of their standard deviation so that a change slightly outside what the training period happened to see does not raise an anomaly.

Each check only appears once its window has enough history to learn from: the 1 Day check after about a week of measurements, the 7 Days check after about a month, and the 28 Days check after about four months.

Scan Assertion and Anomaly Creation

Volumetric Checks are asserted during a Scan Operation just like all other check types and enrichment of volumetric check anomalies is fully supported. This enables full support for custom scheduling of volumetric checks and remediation workflows of volumetric anomalies.

Adaptive Thresholds and Manual Adjustments

Each time data volume is measured for an asset, the system automatically updates the inferred Volumetric Checks.

1.Automatic Threshold Adjustment:

• A newly inferred check pads the observed range of changes by about 1.5 standard deviations.

• On each update the padding is set from how much confidence the platform has in the check, starting as wide as 2 standard deviations and tightening toward 1 as confidence grows, so the thresholds track observed behavior more closely over time.

2.Continuous Learning:

• The system monitors past data and adapts thresholds to detect unusual data volume changes.

3.Why It Matters:

• Helps maintain data integrity by identifying unexpected volume changes.

• Ensures quick detection and response to potential data issues.