Apache Druid

Pivot provides its own semantic layer.  A pivot "data cube" is a logical schema on top of druid, so it is very easy to change out dimensions or measures.

Ah <@U05EWL48JEA> I was about to post a long explanation of the mistakes we made in our architecture.

In larger companies, sometimes your data teams deprecate datasets or even platforms, so the business analysts have to change their reports fairly often.

Even though Druid can support schema migration, the business analysts want to keep their current datasource until the new datasource is fully functional, including loading all the history.  Then they compare before/after versions of the reports, then release the reports to their constituents.