We added a native Zenoh API to ReductStore, meaning it can join a Zenoh network as a peer and persist anything published on a matching key.
You can also query time ranges back out with a normal get().
This is an alternative to other storage backends like InfluxDB, RocksDB, or S3, available in zplugin_storage_manager.
Backends like RocksDB or S3 are good for storing the last state of the network. InfluxDB does have time-series capabilities, but isn’t made for large binary records (images, LiDAR, etc.), and sample attachments don’t map to InfluxDB tags.
ReductStore is a time series object store built for robotics and industrial payloads (ROS messages, logs, images, LiDAR, telemetry).
The mapping between Zenoh’s concepts and ReductStore is basically one to one:
| Zenoh sample | ReductStore record |
|---|---|
| Key | Entry name |
| Timestamp | Timestamp |
| Payload | Payload |
| Attachment | Labels |
Blog: How to Persist Zenoh Data | ReductStore
Disclosure: I work on ReductStore. Happy to answer questions.