A Basic Testbed for JSON Data Processing in SQL Data Servers
Ionuţ Hrubaru, George Talabă, Marin Fotache · 2019
NoSQL data stores, especially the document database systems family (i.e. MongoDB, CouchDB, Couchbase) implemented a long-awaited feature for many software application developers: a flexible schema for storing and processing data. The idea of schemalessness seemed reasonable and the classic trouble of impedance mismatch finally might have had a solution. Main producers of SQL/relational servers reacted by providing support for storing and querying JSON attributes within tables. This paper approaches the problem of comparing JSON data query performance based on the well-established TPC-H benchmark. Normalized schema was converted into JSON format, and all 22 TPC-H queries were translated to process JSON data. Performance of two popular data servers, Oracle and PostgreSQL, was compared. Resulted testbed can be used to assess the performance of any relational/SQL database engines that supports JSON.