MQTT Fundamentals: Publish/Subscribe, Brokers, and Clients

MQTT is my go-to protocol for connecting factory equipment to analytics platforms. Its publish/subscribe model decouples producers and consumers, making it ideal for bridging OT and IT. Here’s how the pieces fit together in real deployments.

1. Publish/Subscribe in Plain Terms

• Publishers send messages to topics (e.g., plant/line1/oven).
• Subscribers receive messages for the topics they care about.
• Broker sits in the middle, authenticating clients, handling routing, and retaining messages when needed.

Unlike request/response protocols, publishers and subscribers never need to know about each other, which simplifies scaling.

2. MQTT Architecture Diagram

flowchart LR
  PLC[PLC / Edge Gateway] --> Broker((MQTT Broker))
  Sensor[Smart Sensor] --> Broker
  Broker --> Dashboard[Dashboards / BI]
  Broker --> CMMS[CMMS]
  Broker --> Twin[Digital Twin]

3. Quality of Service (QoS)

• QoS 0: “At most once” – best effort. Use for non-critical telemetry.
• QoS 1: “At least once” – requires acknowledgment. Use for process values and alarms.
• QoS 2: “Exactly once” – two-phase handshake. Reserve for commands or financial transactions.

Choose QoS per topic based on criticality and network stability.

4. Client Implementation Example

import json
from paho.mqtt.client import Client

client = Client(client_id="edge-line1")
client.tls_set("ca.pem", "edge.pem", "edge.key")
client.username_pw_set("line1", "pass")
client.connect("broker.plant", 8883)

client.publish("plant/line1/temperature", json.dumps({"value": 87.5}), qos=1)
client.subscribe("plant/line1/commands/#", qos=1)

5. Retained Messages and Last Will

• Retained Messages: Broker stores the last message on a topic. Ideal for dashboards that need the latest value immediately.
• Last Will and Testament: Configure clients to announce unexpected disconnects, allowing monitoring systems to alert maintenance.

6. Topic Design Tips

• Use hierarchical topics: plant/line/station/parameter.
• Reserve prefixes for system monitoring (plant/line1/_sys/heartbeat).
• Avoid spaces and keep topics lowercase for consistency.

7. Security Essentials

• Enforce TLS with client certificates.
• Use Access Control Lists (ACLs) to restrict topics per client.
• Rotate credentials and certificates periodically.
• Monitor connection counts and unusual publish rates to detect anomalies.

8. Scaling Strategies

• Deploy clustered brokers (HiveMQ, EMQX, Mosquitto with bridging) for redundancy.
• Use shared subscriptions to load balance consumers.
• Integrate with enterprise IAM for centralized user management.

9. Bridging OT and IT

Pair MQTT with edge gateways that normalize PLC data into structured payloads. Downstream services—dashboards, CMMS, machine learning—consume the same streams without custom point-to-point integrations.

MQTT’s simplicity hides its power. With disciplined topic design, QoS selection, and security controls, it becomes the backbone of resilient industrial IoT architectures.