Does anyone here know at a high level how the database layer works? Specifically, does graphcool create actual tables in the underlying SQL database, is the data stored in a JSON document, does it use an EAV pattern, or something else entirely?

Database per service, table per type, field per field, row per node

nice, thanks!