Telemetry in remote operations: start with the data, not the dashboard
Before thinking about dashboards, you have to solve how the data travels from a site with no stable connectivity. A pragmatic approach with MQTT and local buffering.
· 1 min read · Engineering team · Synapse Group
In distributed operations —field irrigation, pumping stations, remote mining sites— the first question is not "what chart do we want to see?", but "how does the data get here without being lost when the link drops?". The dashboard is the easy part. Transport is the hard part.
The link always drops
A remote operation does not have office-grade connectivity. The satellite or cellular link degrades, cuts out and comes back. If the telemetry system assumes a permanent connection, every outage turns into gaps in the time series and decisions made blind.
The practical consequence: the protocol and the edge architecture matter more than the visualization tool. That is where we start.
MQTT and local buffering
MQTT is lightweight, works over unstable links and decouples who produces the data from who consumes it. The piece that makes the difference is the local buffer: the edge node stores readings while there is no link and ships them when it returns, with QoS 1 to guarantee delivery.
import json
import paho.mqtt.client as mqtt
client = mqtt.Client(client_id="north-pump-01", clean_session=False)
client.username_pw_set("plant", "***")
client.connect("broker.internal.local", 8883)
def publish(reading: dict) -> None:
# QoS 1 + persistent session: the broker retries until delivery is confirmed.
client.publish(
topic="site/north/pump/01",
payload=json.dumps(reading),
qos=1,
retain=True,
)clean_session=False and retain=True are the difference between losing data on
every micro-outage and rebuilding the full series when the link returns.
Only then, the dashboard
With transport solved, visualization becomes a familiar problem. But order matters: a dashboard built on data with gaps gives a false sense of control. The reliable data first; the nice screen second.
This is the kind of decision we make in the field, not on the whiteboard. If you run a distributed operation with unreliable telemetry, let's talk.
- telemetry
- iot
- mqtt