Type 1/2 dimensions


We have some dimensions that can change values like product names etc. When a dimension’s value changes, we’d have to recompute the aggregates, or re-ingest the data. That seems little wasteful of the resources.

So, I’ve done some research on this topic and have some found that there’s not direct support for type 1/2 dimensions, but there are 2 potential solutions (b/c joins are not common on Druid, read https://druid.apache.org/docs/latest/querying/joins.html)--)

Is the above understanding correct ?

Do we always have to provide denormalized data to druid in which dimensions can never change ?

Or, is there a better way of dealing with dimensions that change values ?


Lookups are the most common solution for this.