RTUs and edge gateways at wellheads, block valves, pumping stations and substations publish only what changed, over cellular or satellite. Your SCADA host, historian and on-call team subscribe to one broker — and know at once when a site goes quiet.
No card required
PUBLISH sites/dallas/well-114/telemetry
{"ts":"2026-10-22T06:14:05Z","tubing_psi":412,"casing_psi":188,"flow_mcfd":846,"batt_v":12.7,"cause":"deadband"}QoS 1 · TLS 8883 · PUBACK
A poll-response SCADA host asks every RTU for every register on a fixed cycle, whether anything changed or not. Over cellular and satellite that means data costs, long scan times and gaps when a link drops mid-poll. Every new site adds another VPN tunnel and another point-to-point connection to the host — and when a site goes silent, nobody can tell a dead modem from a quiet well until the next scan fails.
A gateway or MQTT-capable RTU reads the PLC, flow computer and sensors locally over Modbus or DNP3 and turns registers into named values on the site’s topic.
It publishes only when a value leaves its deadband, plus a periodic integrity message — over one outbound TLS connection, with no inbound ports and no polling from the control room.
SCADA/HMI, the historian, leak detection and production accounting subscribe to the same topics. Adding a consumer adds no traffic on the field link.
Threshold rules flag high pressure or low battery and notify on-call staff; Last Will marks a site offline. Setpoints go back down on a command topic only your SCADA host may publish to.
Small MQTT packets, keep-alive tuned to the link, QoS 1 for alarms and persistent sessions, so commands queued while a site was offline are delivered when it reconnects.
Each gateway registers a Last Will. If its connection drops without a clean disconnect, ConnectMQ publishes an offline status — retained, so every dashboard sees the current state immediately.
Every gateway has its own username and password and may publish only under its own site topic; command topics are writable only by your SCADA host. TLS on every connection.
ConnectMQ is a standard MQTT 3.1.1 and MQTT 5 broker, so Sparkplug B edge nodes — which rely on Last Will for their death certificates — can share an instance with simple JSON publishers.
sites/dallas/well-114/telemetry
sites/dallas/well-114/alarms
sites/dallas/well-114/status (retained)
sites/dallas/well-114/cmd/chokeNot at the device level — RTUs and flow computers keep speaking Modbus or DNP3 locally. MQTT replaces the long-haul link: an edge gateway polls on site and publishes changes to the broker, so the WAN carries only report-by-exception traffic instead of full scan cycles.
Yes — MQTT was originally designed for oil pipeline monitoring over satellite. The fixed header is two bytes, sessions survive reconnects, and you choose keep-alive and QoS per message type. If every byte counts, batch or compress payloads at the gateway.
Yes. Sparkplug B is a specification on top of standard MQTT, and ConnectMQ is a standard broker, so Sparkplug edge nodes and host applications such as Ignition connect and use their spBv1.0 topics. ConnectMQ routes the Protobuf payloads unchanged; decoding happens in your host application.
Start on the Free plan — no card required — connect your first devices today and upgrade when you go live.
