standalone cube vs extensible cube (pros/cons/functionality/differences)

Hi, very good question!

I am currently not aware on a specific document that focuses solely on exactly your question. Maybe someone else here is aware and can point us to it. You could have a look at this:

Data Processing and Performance - A comprehensive guide of tables, and design - OneStream Community (onestreamsoftware.com)

Also, if you have access, there is also a Tech Talk under "OneStream Community > News & Views > OneStream Tech Talks: Continue the Conversation > OneStream Tech Talks: Extensibility". I am not sharing the URL here as I am not sure whether that is a publicly available resource. It was published in May 2023.

It depends on the process you would like to create a solution for. There may be valid reasons to create a standalone cube for certain requirements. Generally speaking, setting up a multi-cube solution that leverage extensibility is the recommended option. This gives you the flexibility to create individual data models for e.g. your various legal entities while focusing on the (more granular) reporting data model required at the top. This way, performance is typically much better as the top data unit only stores data relevant at that point, reducing the number of data records that need to be processed. That is, in comparison to a single-cube solution. However, as this is central to each implementation, sufficient time should be taken to discuss this with the corresponding architect.

This is a very short summary from my point of view, the Data Processing and Performance Guide (URL provided above) goes into a lot more detail, overall.