Comparative Analysis of NoSQL and Time Series Databases for Fast Data Acquisition System of Indus-2

Relational databases have been the go-to data logging solution in the scientific community for years. But due to the rapidly changing nature of scientific instruments and the data generated by them, relational databases have begun to fall short of the requirement. In recent years NoSQL and time series databases have come up. These databases have a flexible data model, have horizontal scalability, and are distributed in nature, which helps them handle large volumes of data. This paper provides a comparative study of performance of Cassandra, MongoDB and InfluxDB databases that are used for logging fast data from the data acquisition system of Synchrotron Radiation Source Indus-2. The study uses both the open source synthetic benchmarks and Indus-2 machine parameter data to ascertain the relative performance of the three databases.