Thank you,

Array, struct, variant, maps are advanced data types -- why could they not live in a dimensional model with other primitive data types?

Finding a balance (first image) between OBTs and popcorn analytics is the point of the article; do not default to either unless you fully understand the tradeoffs of either.

1. Joins are expensive - sure, but have you taken advantage or are you aware of the relational platform's join algorithms? I often hear that data professional are not.

2. storage is cheap - it sure is but does that make #3. dimensional modelling dead? Has OBT solved #3 or do you need to solve it every time you query the data? And who solves that? Are your answers the same?

The point is there and it is subtle (intentionally so).

I am happy to answer questions as they arise, this article and a previous one (https://medium.com/the-modern-scientist/data-vault-is-not-a-monolith-3ea2014ffedc) are precursors to the article scheduled for 27 March.

Happy reading!