Related guide: generating-data-vault-from-source-model.

Review the hubs, links, satellites, reference tables, and hub aliases suggested from selected source-model tables, queries, JSON extractions, and pipelines, then apply them to a new or existing .hdv file, or split them into one file per hub, link, and reference table.

How sources are classified

  • Tables (or composed feeds) that share the same primary key become one hub plus one satellite per source.

  • A table whose primary key is made of two or more foreign keys becomes a link (leftover key parts become dependent child keys). Extra columns become a link satellite.

  • Other tables with their own primary key become a hub (and optional satellite). Leftover foreign keys become binary links.

  • Small lookup / code tables can become reference tables (ref_*) instead of hub plus satellite.

  • A self-referencing foreign key becomes a hub alias plus a hierarchy link.

  • Query, JSON, and pipeline cards use the same grain rules. A feed with leftover keys to two or more hubs can become one n-ary transactional link.

  • Isolated tables with no table-to-table relationships are skipped. JSON or pipeline children of those tables can still generate.

Relationship direction is inferred from primary-key and foreign-key columns, so inverted child/parent labels still classify correctly.

Split into separate files

Choose Split into hub and link models when one canvas would be too large.

  • Folder is where the .hdv files are written. A path such as ${PROJECT_HOME}/models is stored on linked hubs.

  • Prefix is optional text in front of each file name. aw becomes aw_hub_customer.hdv. The example line shows the first hub and link file.

  • A project Data Vault model naming scheme (edw-dv-model) formats those file names when one exists.

  • When a file already exists: Skip leaves it unchanged (new satellites are not merged in). Replace overwrites the file.

  • Each hub file contains that hub and its satellites. Each link file contains the link, its link satellites, and linked-table aliases to the hub files. Each reference table gets its own file.

  • Load hub and reference files before link files. Add the files to the resource definition group yourself.

Tips

  • Uncheck a row to skip that object. Edit Vault name before Apply if you want a different name.

  • This builds a raw Data Vault starter. Refine business keys and Business Vault integration afterwards.

  • Use Check model after apply. Generated loads still need catalog feeds (enable Publish unpublished tables when a catalog connection is set).

  • From a Data Vault canvas you can start the same review from the toolbar, the canvas menu, or the coach panel Generate DV button.