Blogs by Category:
Data Vault, dimensional, and data mart design patterns to structure analytics that scale.
September 2, 2026
When a fact references a Type 2 dimension, BimlFlex generates a date range join that picks the dimension row that was true on the fact's effective date, and routes unmatched keys to the unknown member.
August 21, 2026
BimlFlex derives a bridge table's key from the traversal itself: the hub key plus every link key, hashed like any other Data Vault key. Missing legs get the zero key, and the generated load can only add rows, never change them.
August 18, 2026
A PIT load has to know where the last run stopped. BimlFlex derives that watermark from the PIT table itself: MAX of the snapshot dates, a lag-day overlap, and a floor that turns an empty table into a full rebuild.
August 5, 2026
Read the dimension load BimlFlex actually generates: how two row hashes decide what changed, why the end date lands one millisecond or one day back, and how expire and re-insert become a single set-based statement.
July 24, 2026
BimlFlex generates satellite change detection as one hash comparison: what the hashdiff column stores, how the load compares stage against current rows, and how delta collapsing drops rows where nothing changed.
July 16, 2026
Stop hand-writing slowly-changing-dimension logic. BimlFlex generates the Type 1 and Type 2 MERGE behind your dimensions, including date ranges, current flags, and change detection, on every target.