Industrial data connectivity: from register to consumer
Build a trustworthy data path through Modbus, OPC UA, MQTT, Sparkplug and historian storage, with explicit identity and quality.
The outcome of this path is a small data contract that another engineer can implement without guessing units, timestamps, quality or stale behavior. Start at the source rather than at the dashboard.
1. Understand the protocol boundaries
Read Modbus RTU versus TCP and how OPC UA, MQTT and Sparkplug fit. Identify which layer transports values and which gives them shared meaning.
2. Decode and schedule the source
Use the register decoder with a known reference value and review polling and timeout budgets. A fast update is not useful if the datatype or byte order is wrong. Test the unavailable-device path as well as a healthy read.
3. Define the information model
Read OPC UA companion specifications and historian tag naming. Record stable asset identity, engineering units and the meaning of quality. Check what the selected server actually implements.
4. Design message and state behavior
Read MQTT topic and payload contracts, MQTT delivery behavior and Sparkplug basics. Specify late-subscriber, reconnect, duplicate and stale-source behavior rather than assuming QoS solves each one.
5. Preserve useful history
Read historian sampling and deadband. Estimate raw storage from tag count and sample rate, then test a known transient against the stored trace. Distinguish source acquisition from archive filtering.
Deliverable
Complete the industrial data-contract worksheet for one signal. Include a named owner, revision, timestamp boundary, quality mapping and consumer behavior when the source stops. Pair the contract with the OT reliability and security path before a production rollout.