SEER is an online tool to evaluate the performance of time series database systems on large datasets. The tool builds upon our TSM-Bench benchmark TSM-Bench: Benchmarking Time Series Database Systems for Monitoring Applications, PVLDB'23. SEER compares seven Time Series Database Systems (TSDBs) using a mixed set of workloads. It implements a novel data generation method that augments seed real-world time series datasets, enabling realistic and scalable benchmarking.
- List of benchmarked systems: ClickHouse, Druid, eXtremeDB*, InfluxDB, MonetDB, QuestDB, TimescaleDB.
- SEER evaluates bulk-loading, storage performance, offline/online query performance, and the impact of time series features on compression.
- The tool uses two datasets for the evaluation: D-LONG [d1] and D-MULTI [d2]. The evaluated datasets can be found here.
- *Note: Due to license restrictions, we can only share the evaluation version of extremeDB. The results between the benchmarked and the public version might diverge.
SEER was created at the eXascale Infolab, a research group at the University of Fribourg, Switzerland.
- Ubuntu 20 (including Ubuntu derivatives, e.g., Xubuntu)
- Clone this repository
- Install Docker and Docker-Compose if not already installed
sh setup/install_docker.sh- Build the app and start it
docker-compose up -d --build- After setting up the tool, you can access it by opening http://localhost:12007 in your browser. If the tool does not launch, please review the docker installation
sh setup/init_seer.sh
sh setup/migrate_query_data.sh- Luca Althaus
- Mourad Khayati (mkhayati@exascale.info)