ModelarDB is a modular model-based time series management system that interfaces a query engine and a data store with ModelarDB Core. This Core is a self-contained, adaptive, and highly extensible Java library for automatic online compression and efficient aggregation of time series. ModelarDB is designed for Unix-like operating systems and is tested on Linux.
This repository primarily contains the static source code for the legacy JVM-based versions of ModelarDB documented in the papers listed below. A re-design and re-implementation in Rust is currently being developed as an open-source project at ModelarDB. Thus, this Rust-based version of ModelarDB is under active development, while the legacy JVM-based version of ModelarDB is not.
- Install a Java Development Kit.*
- Install the Scala Build Tool (sbt).
- Clone the ModelarDB Git repository.
- Start
sbtin the root of the repository and run one of the following commands:
compilecompiles ModelarDB to Java class files.packagecompiles ModelarDB to a Jar file.assemblycompiles ModelarDB to an uber Jar file.**runexecutes ModelarDB's main method.***
* OpenJDK 11 and Oracle's Java SE Development Kit 11 have been tested.
** To execute ModelarDB on an existing Apache Spark cluster, an uber Jar must be created to ensure the necessary dependencies are included in a single Jar file.
*** If sbt run is executed directly from the command-line, then the run
command and the arguments must be surrounded by quotes to pass the arguments to
ModelarDB: sbt 'run arguments'.
ModelarDB requires that a configuration file is available at
$HOME/.modelardb.conf or is passed as the first command-line argument. This
file must specify the query processing engine and data store to use. An example
configuration is included as part of this repository.
The research leading to ModelarDB is documented in several papers. If you use ModelarDB in academia, please cite the relevant paper(s) below.
Why Model-Based Lossy Compression is Great for Wind Turbine Analytics
by Søren Kejser Jensen, Christian Thomsen, Torben Bach Pedersen, Carlos Enrique Muñiz-Cuza, and Abduvoris Abduvakhobov
in The Proceedings of ICDE, 5667-5668, 2024
Links: IEEE, Slides
Time Series Management Systems: A 2022 Survey
by Søren Kejser Jensen, Torben Bach Pedersen, and Christian Thomsen
in Data Series Management and Analytics (Forthcoming)
Links: AAU (preprint)
ModelarDB: Integrated Model-Based Management of Time Series from Edge to Cloud
by Søren Kejser Jensen, Christian Thomsen, and Torben Bach Pedersen
in Transactions on Large-Scale Data- and Knowledge-Centered Systems LIII, 1-33, 2023
Links: Springer
Scalable Model-Based Management of Correlated Dimensional Time Series in ModelarDB+
by Søren Kejser Jensen, Torben Bach Pedersen, and Christian Thomsen
in The Proceedings of ICDE, 1380-1391, 2021
Links: IEEE, arXiv (preprint), Slides
Model-Based Time Series Management at Scale
by Søren Kejser Jensen
PhD Thesis, The Technical Faculty of IT and Design, Aalborg University, 2019
Links: AAU, Slides
Demonstration of ModelarDB: Model-Based Management of Dimensional Time Series
by Søren Kejser Jensen, Torben Bach Pedersen, and Christian Thomsen
in The Proceedings of SIGMOD, 1933-1936, 2019
Links: ACM
ModelarDB: Modular Model-Based Time Series Management with Spark and Cassandra
by Søren Kejser Jensen, Torben Bach Pedersen, and Christian Thomsen
in The Proceedings of the VLDB Endowment, 11(11): 1688-1701, 2018
Links: PVLDB, Slides
Time Series Management Systems: A Survey
by Søren Kejser Jensen, Torben Bach Pedersen, and Christian Thomsen
in IEEE Transactions on Knowledge and Data Engineering, 29(11): 2581–2600, 2017
Links: IEEE, arXiv (preprint)
The research leading to ModelarDB has also been presented at multiple events. The slides used are available below:
ModelarDB: Analytics of High-Frequency Time Series Across Edge, Cloud, and Client
by Christian Thomsen and Søren Kejser Jensen
at Danish Digitalization, Data Science and AI, 2024
Links: Event, Slides
Model-based storage and management of massive sensor time series
by Christian Thomsen
at Digital Tech Summit, 2021
Links: Event, Slides
Extreme-Scale Model-Based Time Series Management with ModelarDB
by Torben Bach Pedersen as a Keynote Speaker
at the 32nd International Conference on Database and Expert Systems Applications, 2021
Links: Event, Slides
Extreme-Scale Model-Based Time Series Management with ModelarDB
by Torben Bach Pedersen as a Keynote Speaker
at the 28th International Symposium on Temporal Representation and Reasoning, 2021
Links: Event, Abstract, Slides
Extreme-Scale Model-Based Time Series Management with ModelarDB
by Torben Bach Pedersen as a Keynote Speaker
at the 10th International Conference on Model and Data Engineering, 2021
Links: Event, Slides
Model-Based Management of Correlated Dimensional Time Series
by Søren Kejser Jensen
at Dagstuhl Seminar 19282 (Data Series Management), 2019
Links: Event, Rapport, Slides
Effektive metoder til at gemme og forespørge på store mængder tidsseriedata
by Christian Thomsen
at GrowAAL, 2019
Links: Event, Slides
ModelarDB is licensed under version 2.0 of the Apache License and a copy of the license is bundled with the program.