With JSON Relational Duality, developers can leverage JSON documents for data access while using the highly efficient relational data storage model, without compromising simplicity or efficiency.

Including columns of JSON data type in tables that underlie a duality view lets applications add and delete fields, and change the types of field values, in the documents supported by the view. The stored JSON data can be schemaless or JSON Schema-based (to enforce particular types of values).

When you define a duality view, you can declaratively choose the kind and degree of schema flexibility you want, for particular document parts or whole documents.

A duality-view flex column stores (in an underlying table) JSON objects whose fields aren't predefined: they're not mapped individually to specific underlying columns. Unrecognized fields of an object in a document you insert or update are automatically added to the flex column for that object's underlying table.

You can thus add fields to the document object produced by a duality view with a flex column underlying that object, without redefining the duality view. This provides another kind of schema flexibility to a duality view, and to the documents it supports. If a given underlying table has no column identified in the view as flex, then new fields are not automatically added to the object produced by that table. Add flex columns where you want this particular kind of flexibility.