/caveat i know nothing about how mongo is storing data and haven’t used it since 2010

it doesn’t really have to approach the schemas that way. one would think it would be optimized for repeat schema in the same way one might create the schema definition and then reference it in packing and unpacking the data. seems like if there was schema overhead taking up storage unnecessarily that could be optimized relatively easily.

having schema references also might be a good management tool to understand which records vary potentially due to an application evolving it’s needs.