Skip to main content
Navigation

Volume Monitoring

Track row counts and detect unexpected changes in data volume.

Volume monitoring helps you catch issues that freshness checks miss — like a pipeline that runs on time but loads zero rows, or a table that unexpectedly doubles in size.

FreshGuard offers two types of volume monitoring.

Volume Threshold Starter+

A simple, static check: alert when the row count in a table goes above or below a value you define.

Configuration

  • Minimum rows — Alert if the table has fewer rows than this number.
  • Maximum rows — Alert if the table has more rows than this number.

You can set one or both thresholds depending on what you want to catch.

Use Cases

  • Empty table detection — Set minimum to 1 to catch tables that are accidentally truncated.
  • Row count sanity check — If your daily_orders table normally has 10,000–50,000 rows, set a minimum of 5,000 and maximum of 100,000 to catch major deviations.
  • Storage growth — Set a maximum to be alerted before a table grows beyond expected limits.

Info

Volume Threshold checks count the total number of rows in the table at the time of the check. They don’t track changes between checks — for that, use Volume Anomaly.

Volume Anomaly Professional+

ML-powered anomaly detection that learns your table’s normal volume patterns and alerts when the current count deviates significantly.

How It Works

  1. FreshGuard records the row count at each check interval.
  2. After enough data points, the system builds a baseline model of your table’s normal volume pattern (accounting for daily and weekly seasonality).
  3. On each subsequent check, the current row count is compared against the expected range.
  4. If the count falls outside the expected range, an anomaly alert is triggered.

When to Use It

Volume Anomaly is ideal for tables where the “normal” row count varies by time of day or day of week — patterns that are hard to capture with a static threshold. Examples:

  • Event tables that see higher volume during business hours.
  • Transaction tables with weekly seasonality (lower on weekends).
  • Staging tables where the expected volume depends on upstream activity.

Tip

Volume Anomaly needs at least a few days of historical data to build an accurate baseline. During the learning period, you’ll see a “Learning” status on the rule.

Combining with Freshness Checks

Volume monitoring and freshness checks are complementary. A common setup for a critical table:

  1. Freshness Check (Free) — Catch when data stops arriving entirely.
  2. Volume Threshold (Starter) — Catch when the pipeline runs but loads zero or very few rows.
  3. Volume Anomaly (Professional) — Catch subtle drops that a static threshold would miss.

Each rule type runs independently and can alert through different destinations if needed.