Comparison
Data classificationvsPersonally identifiable information
Data classification
every column is tagged as public, internal or restricted, and the access rules are generated from the tags rather than written per table.
Labelling data by sensitivity so that policies can be applied by label rather than table by table. It is what makes governance scale past a few dozen tables, because a rule expressed over a tag automatically covers next month's new table. Its weak point is coverage: an untagged column defaults to whatever the platform's default is, so the default has to be the restrictive one.
Full entry →Personally identifiable information
a free-text notes column turned out to contain names and phone numbers, and it had been copied into six downstream tables.
Data that identifies a person directly or in combination with other fields. What makes it hard in a warehouse is not the obvious columns but the propagation: it arrives in unstructured fields, gets joined, copied into marts and exported, and a control applied only at the source protects none of the copies. Classification therefore has to travel with lineage or it is decorative.
Full entry →