Glossary · IIoT, data and AI
Time-series database
Also known as: TSDB
German: Zeitreihendatenbank
In data management, a time-series database is a database optimized for storing and querying values indexed by time, such as sensor readings, machine states and events, with efficient compression, high write rates, retention policies and time-based aggregation functions.
- IIoT
In one sentence
A time-series database is optimized for storing and querying time-stamped values such as sensor readings at high write rates.
Example
Temperatures from 2,000 tags are written once per second to a time-series database, which keeps raw data for 90 days and hourly averages for ten years.
How it applies
- Engineering: Choose sampling or change-based recording per signal, define retention and downsampling, and store each series with tags for asset, unit and source.
- Operation: Time stamps must be consistent. Clock synchronization and clear time zone handling (ideally UTC) prevent misleading comparisons between signals.
- Documentation: Document the naming scheme, retention rules and the meaning of each series. Without this, the database fills with values nobody can interpret.
Time-series database vs. historian
A Historian is an industrial software product that combines time-series storage with OT connectivity, compression, visualization and often validated audit functions. A general time-series database is an IT component that offers storage and queries but needs connectors and tools around it. Operational dashboards, Condition monitoring and analytics tools usually read directly from the time-series database.