Field Health

The Field Health checks available for each column type, and what they have in common

Field Health monitors watch a single column and test every row in scope against a rule. Each run reports how many rows failed, and compares that count against the Tolerance you set, so you decide how many bad rows the column can carry before the monitor raises an incident.

The check you can use depends on the column's type.

Numeric columns

Check	What it tests	Page
Average, Minimum, Maximum, Sum	One aggregate statistic for the column falls inside a range	Numeric checks
Numerical Range	Every individual value falls inside a range	Numerical range
Zero	The column carries few enough zero values	Condition checks

String columns

Check	What it tests	Page
String Length	Text length falls inside a range	String checks
Regex Match, Regex Not Match	Values match, or avoid, a pattern	String checks
Value In, Value Not In	Values belong to, or stay out of, a defined set	Value set checks

Date and timestamp columns

Check	What it tests	Page
Is Before, Is After	Values fall before or after a point in time	Temporal checks

Any column type

Check	What it tests	Page
Null	The column carries few enough empty values	Condition checks
Unique	Values repeat less often than your Tolerance allows	Condition checks

Numerical Range and String Length use a Threshold and a Tolerance together: the Threshold defines which rows fail, the Tolerance defines how many failing rows the run accepts. The Condition checks come with a fixed rule built in, so Tolerance is the only lever you set.

Running one manually

Field Health monitors support a manual schedule. A manual run scans the full table, so factor that cost in on large tables before choosing it over a schedule.