Pushing the Limits of Tango Archiving System using PostgreSQL and Time Series Databases
Reynald Bourtembourg, Guifré Cuní, Matteo Di Carlo, George Fatkin, James, Stuart, Lorenzo Pivetta, Pons, Jean-Luc, Sergi Rubio-Manrique, C. Scafuri, Graziano Scalamera, A. I. Senchenko, Ситнов Владимир Евгеньевич, Giacomo Strangolino, Pascal Verdier, Lucio Zambon · HAL (Le Centre pour la Communication Scientifique Directe) · 2020
The Tango HDB++ project is a high performance event-driven archiving system which stores data with micro-second resolution timestamps, using archivers written in C++. HDB++ supports MySQL/MariaDB and Apache Cassandra backends and has been recently extended to support PostgreSQL and TimescaleDB*, a time-series PostgreSQL extension. The PostgreSQL backend has enabled efficient multi-dimensional data storage in a relational database. Time series databases are ideal for archiving and can take advantage of the fact that data inserted do not change. TimescaleDB has pushed the performance of HDB++ to new limits. The paper will present the benchmarking tools that have been developed to compare the performance of different backends and the extension of HDB++ to support TimescaleDB for insertion and extraction. A comparison of the different supported back-ends will be presented.