Abstract
Being a simple semi-structured data model, JSON has been widely accepted as a simple way to store, query, modify, and exchange data among applications. In comparison with schema-oriented relational data model to store, query, and update application data, JSON data model has the advantage of being self-contained, free from schema evolution issues, and flexible enough to enable agile style development paradigm. Therefore, during the last 5 years, SQL/JSON standard has been established as foundation for managing JSON data in SQL standard and there have been JSON functionalities added into RDBMS products to support SQL/JSON standard to various degree. In this chapter, the authors will analyze the strength and weakness of using JSON as the data model to manage data for applications. For use cases where JSON data model is ideal, they present the design approaches to store, index, query, and update JSON in the kernel of RDBMS to support SQL/JSON standard defined operations effectively and efficiently.
Key Terms in this Chapter
OSON: Oracle optimized binary JSON encoding format that is capable of running jump navigation to drastically support SQL/JSON path language evaluation and is capable of doing piece-wise update of OSON bytes to avoid full document replacement.
JSON Search Index: A schema-agnostic indexing approach over a JSON collection that does not make any assumption of specific schema JSON documents. It speeds up ad-hoc SQL/JSON path value and keyword search queries over JSON collection.
JSON DataGuide: A summary of all possible JSON paths and leaf value domain type for a collection of JSON documents to assist formulation of query to the JSON collection. It is conceptually the computed schema of the JSON collection.