Dimension table

Table of Contents

Edit a Kimball dimension on a .hdm canvas.

General

Name, physical table, description, and staging source (none / logical, SQL, pipeline, record definition, or date generator).

Load strategy is derived from the attribute SCD policies. When every attribute is Type 1, the field offers two choices:

Type 1 — overwrite writes every source row with Insert/Update.
Type 1 — skip unchanged compares the source with the dimension table (Merge rows (diff)) and sends only new and changed rows to Insert/Update. The load timestamp and load cycle id are not compared. Identical rows keep the values already stored. Rows present in the table and absent from the source are not deleted.

Type 2, Type 3, and hybrid dimensions show the derived strategy and ignore a stored skip-unchanged choice until every attribute is Type 1 again.

Surrogate key — Strategy (none, auto-increment, source field such as a DV hash) and column name.
Natural keys — Business identity used for lookup.
Attributes — Type 1 / Type 2 / hybrid columns and SCD policy.
Outriggers — Optional snowflake references.

Edit field… (or double-click a row) opens the field editor for Markdown notes and requirements.

Configure how staging rows are read for generated Dimensional Update pipelines.

None (logical) — sketch the dimension without a load source. Check model warns; Dimensional Update skips the table.
Other types still require a complete source before a load.