Forum Discussion
LakeHouse Models - Has this changed?
Hi, I recently built a LakeHouse based semantic model that uses tables from two different LakeHouses. One is for my primary datasource and the other for master data tables like locations and a da...
In general a relationship based on datetime is undesirable due to the inherently high cardinality. Usual recommendation is to break it up into separate date and time dimensions.
Originally I planned to use straight date fields but found (perhaps a bug) that any date fields when loaded into a LakeHouse table would become a datetime field. To fix I made all dates into datetimes with 12:00am as the time.
yes, to make it even more confusing - There is not actually a "date" or "time" column type in DAX/VertiPaq. Behind the scenes everything is always datetime. The "date" values are datetimes pegged at 12:00:00am.
The cardinality issue still holds true though.