Sure. But why use Mongo for that, instead of a PostgreSQL table with a JSON column for the data, and perhaps a denormalized version identifier so your applicati

Sure. But why use Mongo for that, instead of a PostgreSQL table with a JSON column for the data, and perhaps a denormalized version identifier so your application code knows what to do with the format of a given row's data field? I can see a speed argument, but I can't see how that militates for pure Mongo, instead of Mongo as a cache in front of something that provides a reliably (not "eventually") consistent backing store.