datarevisjoner

Datarevisjoner refers to the recording of changes to data over time, creating a history of edits and versions to ensure traceability, accountability and data integrity. They are a key aspect of data governance in transactional systems, data warehouses and data lakes. A datarevision captures what changed, when it changed and who approved or performed the change. Revisions can be stored as audit logs, dedicated history tables, temporal (time-point) tables, or as event streams, enabling retrieval of past states and rollback if needed.

Common approaches include slowly changing dimensions type 2 in data warehousing, which preserves previous versions of a record with validity intervals, and temporal databases that support bi-temporal queries. In modern data architectures, append-only logs (such as event stores) or versioned datasets (Delta Lake, Apache Hudi, Apache Iceberg) are used to track revisions across large datasets.

Governance considerations include access controls, approval workflows, retention policies, and privacy requirements. Historical data may require masking or selective deletion to comply with data protection laws. Metadata management and data lineage are essential to understand the source, purpose, and impact of revisions.

Benefits include improved data quality, reproducibility of analyses, auditability for compliance, and safer rollback in case of errors. Challenges include storage growth, performance overhead, and balancing transparency with privacy. Organizations often establish policies for how long revisions are kept, how they are labeled, and how they are accessed.

See also: data provenance, data lineage, data quality, version control for data. Datarevisjoner thus underpin reliable data management and informed decision-making.