MQTT Sparkplug B: Speaking the Lingua Franca of Industrial IoT

We’ve all heard that MQTT is the “de facto” standard for IoT. Ideally, it is lightweight, report-by-exception, and decouples devices from applications.

But in raw MQTT, if a sensor publishes {"val": 24.5} to topic factory/sensor1, the consumer has no idea if that’s degrees Celsius, PSI, or the lunch break duration. Worse, if the device dies, the broker is silent.

Enter Sparkplug B. It is the specification that makes MQTT “grown-up” enough for mission-critical industrial control.

The “Wild West” of Raw MQTT

I once audited a SCADA system where five different PLCs published JSON payloads with five different schema formats. The engineering team spent 60% of their time writing parsers and maintenance scripts.

Raw MQTT lacks:

1. State Management: You don’t know if a device is offline or just quiet.
2. Data Typing: Is “1” a boolean true or an integer 1?
3. Discovery: You have to manually map every single tag.

Why Sparkplug B Changes the Game

Sparkplug B solves these specifically for the OT world:

1. Birth & Death Certificates: When a node connects, it publishes a “Birth” message explaining exactly what tags it has. When it drops offline gracefully or ungracefully, the broker issues a “Death” certificate. Your SCADA immediately creates a “Bad Quality” overlay.
2. Defined Payload: It uses Google Protobufs. It’s binary, compressed, and strictly typed. No more JSON parsing errors.
3. Report by Exception (RBE): It only moves data when it changes. We’ve seen bandwidth reductions of 90% compared to polling protocols like Modbus TCP.

Implementing Sparkplug in 2026

The ecosystem has matured significantly.

• PLCs: Modern Opto22 and WAGO controllers speak Sparkplug out of the box.
• Gateways: We use HiveMQ Edge or Kepware to translate legacy Modbus traffic into Sparkplug B streams.
• SCADA: Ignition was the pioneer, but now nearly every major platform consumes it natively.

The Payoff

By adopting Sparkplug B, you stop building point-to-point connections. You build a Unified Namespace where any new application—be it an MES, a maintenance AI, or a dashboard—can plug in and immediately discover the entire plant model.

Is your data pipeline a mess of custom JSON parsers?

I help companies standardize their edge communications to build robust, scalable IIoT infrastructures. Let’s clean up your architecture.