Oracle Globally Distributed Database allows JSON documents to scale to massive data and transactions volume, provide fault isolation, and support data sovereignty. Oracle Database has support for native JSON objects. Applications can interact with the distributed database using the SODA (Simple Oracle Document Access) API, which allows you to access data using JSON document attributes.

In Oracle Database, JSON documents can be stored in a database table. The database tables act as JSON collections, and each row is a JSON document. JSON documents are stored in the database as type JSON, which is backed by a highly optimized binary JSON format called OSON.

Although Oracle provides support for JSON operators to create, work with, and retrieve JSON documents, the SODA interface is also supported. SODA provides a more intuitive interface for working with JSON documents.

SODA is an API for NoSQL-style JSON (and not only JSON) document collections in Oracle Database. Using SODA APIs, application can perform CRUD operations on documents in collections. Collections are backed by regular Oracle tables (or views).

Typically, to create a collection, one would use SODA API. That creates the underlying table backing the collection. In order to create a sharded collection, however, a shared table has to be created first. Then, a collection can be created on top of a sharded table, by using a mapped collection feature of SODA.

Working with JSON documents in a distributed database introduces the notion of a sharding key. JSON documents are distributed to the individual database table shards according to the sharding key. The sharding key can either be a field from within the JSON document or an external column such as the ID assigned by the SODA API.

For further reading about JSON and SODA, see JSON in Oracle Database and Overview of SODA.

The topics that follow provide details about how to shard JSON objects in Oracle Database. The high level steps are:

• Deploy a distributed database
• Identify a sharding key that the application can use to fetch data
• Define a data store for JSON in Oracle Database by creating sharded tables
• Map the sharded table with SODA

Then life cycle management tasks detailed are:

• Add documents to the sharded JSON collection in the application
• Fetch document data from the sharded JSON collection in the application