You deployed ten robots. The system worked. Now you have a hundred, and your engineering team is complaining about things that didn't exist six months ago: storage filling up on edge devices, cloud bills growing faster than the fleet, retraining cycles that take weeks instead of days, and incident investigations lack the data they need.

This is a predictable consequence of crossing a scaling threshold that almost every AMR operator hits.

MQTT is a common choice for the communication stack in IoT and robotics applications because it is lightweight and easy to integrate. But many of those applications do not send only small JSON telemetry messages. They also publish JPEG frames, vibration waveforms, audio clips, protobuf messages, and other binary payloads that need to be stored and queried later.

This is where a regular MQTT broker or a traditional time-series database starts to fall short. Brokers are designed for message delivery, not long-term historical storage, and many databases either expect structured numeric fields or make it hard to keep large binary records tied to accurate timestamps.

In this tutorial, we will use ReductBridge to subscribe to MQTT topics and write the raw binary payloads into ReductStore with a time index. This lets you keep camera frames and sensor payloads as they are, while still querying them by time range, labels, and entry name for replay, debugging, and offline analysis.

The CRA Deadline Every German Robot Operator Must Face

The EU Cyber Resilience Act (Regulation (EU) 2024/2847) is the “GDPR for connected products.” It entered into force on 10 December 2024, with critical milestones approaching fast:

• 11 September 2026: Mandatory reporting of actively exploited vulnerabilities and severe incidents (24-hour early warning, 72-hour full notification).
• 11 December 2027: Full compliance — Security by Design, lifecycle support (minimum 5 years), technical documentation, and CE marking.

For robotics fleets (AMRs, cobots, autonomous systems, and ROS 2-based platforms) the stakes are particularly high. These systems are “products with digital elements” (often Class II or critical), generating massive multimodal data streams (camera feeds, LiDAR, IMU, logs, ROS bags) under real production constraints: intermittent connectivity, edge hardware limits, and high physical safety risks.

Generic storage solutions force painful trade-offs: either accept data loss and compliance gaps, or accept exploding costs and slow performance. ReductStore eliminates this trade-off.