Documentation for OpenTSDB 3.0

Welcome to OpenTSDB 3.0, the scalable, distributed time series database.


Version 3.0 is still under active development and many features may change before it’s release. This documention should reflect the current state of the code at any time but if you see something outdated, please let us know or give us a PR.

3.0 is a massive change for OpenTSDB with improved query performance and flexibility yet it maintains backwards compatibility as much as possible with previous versions. Data written to HBase or Bigtable from previous versions can still be read (though annotations and histograms won’t work quite yet, coming soon).

The HTTP API is backwards compatible with 2.x via /api/query but note that 1.x querying has been dropped.

And the Telnet style API has not been reimplemented yet. You can write data via the HTTP API for now.

Indices and tables


We’re also in the process of re-writing the 3.0 Java API a bit. Right now data is handled in a memory efficient manner but it doesn’t take advantage of CPU caching and is rather slow. We’re adding a push model that will allow for vectors similar to other TSDBs while also allowing for precise raw data processing.