COMPARATIVE ANALYSIS OF THE SQL AND NOSQL DATABASE PRODUCTIVITY

Abstract

A brief analysis of SQL and NoSQL database functions was performed in this work, their main differences were cited. To date, there are two most common types of data management systems: relational databases and NoSQL. There is a huge variety of data models and API (Application Programming Interface) requests for NoSQL. In particular, for comparison, Apache Cassandra, DynamoDB, MongoDB were selected. The data model and functionality of Apache Cassandra are similar to other scalable repositories. Updates and grouping of columns are cached in RAM and then reset to disk. The main purpose of the work was to compare the performance of relational SQL databases and NoSQL databases, for example, PostgreSQL, MySQL, Apach Cassandra, MongoDB, Amazon DynamoDB. A separate software product was developed for testing the performance. The main subject of the study is the performance of these systems basic operations. Performance results for each of them were obtained using testing system adapted for research purposes. The developed system of testing allowed testing the speed of complex analytical operations, making additional settings, using the large amount of data. The system has been expanded for testing the extended set of operations over a data plan that contains links between tables. This testing system contains a set of ready-made loads that cover the main aspects of the operation and support user-generated load. The testing system received data on the performance of the presented database management systems for a set of different queries. For performance analysis, the system response time (the time between the beginning of the request and the response) was measured on request. Two types of indicators were compared - the average response to completed operations and detailed analysis. The obtained data was presented in the form of diagrams, and by them was made a conclusion on the performance of SQL and NoSQL databases. The choice of databases should be based as much on the type of tasks to be solved and should also take into account the amount of data and the system response time.