When you connect to ADL, the data is "schema-less" (ie there are no field names on the file columns.). Data Lake has a way of presenting multiple files (i.e the data files - there could be multiple.., the json file containing the field definitions...