Comparison of JSON and XML Data Formats in Document Stored NoSql Database Replication Processes

Abstract

The current trend of solutions in storing large amounts of data is using the NoSQL Database. A document stored is one type of NoSQL database that uses the JavaScript Object Notation (JSON) and eXtensible Markup Language (XML) data formats in data storage. High Availability database is significant to support cloud-based applications and services. Replication is one solution to maintain the consistency of source data and target data. This study aims to determine the performance of JSON and XML data formats in the document stored NoSQL Database replication process. In this study, ArangoDB, RethinkDB, and MongoDB were chosen for use in the trial process of replication from master-server to slave-server with two different data formats, JSON and XML. Data transfer, CPU usage, memory usage, and execution time are measured in each trial. Based on research and experiments that have been carried out, the JSON data format consumes bandwidth with an average value smaller than the XML data format; this occurs in MongoDB, CouchDB, and RethinkDB. In CPU usage, JSON data format, on average, consumes less CPU compared to the XML data format. This is the case with MongoDB. While on CouchDB and RethinkDB, the average CPU usage for XML and JSON data formats does not show a significant difference. The average memory usage for the JSON data format is smaller than the XML data format. The average execution time of the XML data format a little faster than the JSON data format.