Alexey Timin

A robot or production line writes sensor data continuously. Eventually, a machine crashes, a disk fails, or a bad deployment corrupts an index. The real question is how much data loss and downtime you can afford when that happens.

The Disaster Recovery guide is the product reference: it describes the available mechanisms and their settings. This post focuses on choosing and deploying them. It covers cold backup, hot backup with the CLI, hot standby, and read-only replicas, including the downtime each requires and when each is appropriate for a robot or plant that cannot stop.

Plenty of equipment on the shop floor will never speak MQTT or OPC-UA. It predates both. What it has is a command line: a vendor tool that prints status, a diagnostic command on a serial console, or a script the previous engineer left behind. The data exists. There is just no protocol to integrate.

Reduct Bridge has a shell input for exactly this case. You point it at a command, set an interval, and the bridge runs it on schedule. Each line the command prints becomes a time-indexed record in ReductStore. Labels come from regex capture groups in the output or from static key-value pairs in the config. Wrap whatever the vendor tool prints, put it on an interval, label it, and it lands time-indexed beside everything else.

This post walks through capturing telemetry from a legacy test chamber that only exposes a chamber-status command.

We originally built ReductStore to store data on edge devices and read it back by time interval. But a device only has so much disk space. With cameras and sensors writing all the time, we used a FIFO quota to remove the oldest records and make room for new ones. That left us with another problem: how to get the data off the device before it was deleted.

At first, transferring data from an edge device to central storage was a manual job. We used the CLI client or scripts and copied data when somebody remembered to do it. It was always problematic because data collection never stopped. Sometimes we had only two or three days to copy it before it was gone, and a temporary network problem or a missed run could make that window disappear completely.

Automatic replication was the next logical step. The edge is producing a stream of new data, while the central store has more capacity and can have different criteria for what it keeps. We needed to forward new records and configure source-side filters for each destination. Keeping two identical replicas was not the goal.

We also had to keep in mind that the source is usually on the edge. Very often it has no public IP address, and its network connection can be unstable, slow, or absent for long periods. Replication therefore could not make local ingestion wait for the central store. The device still needed to accept data first and deliver it later when a connection became available.