Define the data contract
Mapping a Modbus register to an MQTT topic is only the beginning. Define units, scale, source, collection time and data quality. Without that context, a number can reach the dashboard and still represent the wrong information.
Separate telemetry, events and commands. Each flow has different frequency, priority and acknowledgment needs. Version the contract so the device and the application can evolve without requiring a simultaneous fleet-wide update.
Decide what works without the internet
A dependable architecture defines which decisions remain local and which depend on the platform. Process control and functions with response-time requirements should be assessed separately from historical data uploads and remote visualization.
Implement store-and-forward where needed, with storage limits and discard rules. Identify messages and events so the backend can recognize duplicates. Test recovery of both equipment and platform connections, including extended outages.
From equipment to application
- 01
Equipment
Sensors and machines. Modbus, CAN and local interfaces.
- 02
Gateway
Data collection, context and network connectivity.
- 03
Platform
MQTT, APIs, identity and device history.
- 04
Applications
Alarms, maintenance and operational decisions.
At the edge: local decisions and storage keep the integration useful during internet outages.
Make the integration operable
Alongside process data, monitor gateway health: firmware version, connection quality, storage usage and last communication with each connected device. These signals help separate sensor issues from network failures.
Remote configuration, per-device credentials and access control belong in the architecture. Record changes and commands with enough context to investigate who did what and on which device.
Value for the business and the engineering team
For US teams, explicit interface contracts and English documentation make the gateway easier to integrate with applications, system integrators and support processes. In Brazil, connecting installed machines can support gradual digitalization while preserving existing investments.
Commercial value emerges when the data informs a decision: identifying downtime, tracking alarms or adjusting operations. Start with a clear use case and a metric the business already understands, then expand the integration.
