Grafana for IoT: real-time dashboards and alerting explained

A sensor reading, on its own, is just a number with a timestamp. Grafana is the piece of the stack that turns thousands of those rows, arriving continuously from every sensor on every site, into something a person can actually look at: a line trending upward, a gauge in the red, a map with sites colour-coded by status. It is an open-source visualisation and alerting tool built specifically for time-series data, which makes it a natural fit for IoT.

Dashboards: panels wired to a data source

A Grafana dashboard is a grid of panels, each one a query against a data source (in Nodenet’s case, the time-series database every sensor reading is written to) rendered as a chart, gauge, table, or map. Building one is mostly a matter of writing a query, "show me vibration readings for site 12 over the last 24 hours", and picking how to display it. Panels refresh on a timer, so a dashboard left open on a site office screen keeps showing current conditions without anyone touching it.

Alerting: rules that watch the data for you

A dashboard someone has to remember to check is only half useful. Grafana’s alerting engine lets you define a rule, "notify me if noise readings exceed 85dB for more than five minutes", that evaluates continuously in the background and fires a notification the moment the condition is met, without anyone watching the screen. Alerts can route to email, SMS, or a webhook that hands off to an automation flow for a more elaborate response.

Why time-series data specifically

General-purpose business intelligence tools can technically chart a number over time, but they are built around infrequent, ad-hoc reports, not a firehose of readings arriving every few seconds from hundreds of devices. Grafana is built for exactly that shape of data: it queries efficiently against large time ranges, handles gaps and irregular reporting intervals gracefully, and is designed to stay responsive as a dashboard someone leaves open all day, not a report generated once and forgotten.

A real example: one dashboard, five sites, colour-coded by risk

Picture a portfolio of five active sites, each with vibration, noise, and dust sensors reporting over a mix of MQTT and LoRaWAN. A single Grafana dashboard can show all five as a row of panels, each one green, amber, or red depending on how close current readings sit to that site’s threshold, so a manager can see the whole portfolio’s status in one glance rather than logging into five separate systems. Underneath, every panel is querying the same normalised data, regardless of which protocol a given sensor originally reported over.

Roles: who can see what, and who can change it

Not everyone who looks at a dashboard should be able to edit it or change alert rules. Grafana supports role-based access, typically Admin (full control), Editor (can build and adjust dashboards), and Viewer (read-only), so a site supervisor can watch live conditions without being able to accidentally delete a panel or silence an alert.

• A dashboard is a grid of panels, each a query against your time-series data rendered as a chart, gauge, or table
• Alerting rules evaluate continuously and fire the moment a condition is met, no manual watching required
• Grafana is purpose-built for time-series data, unlike general business intelligence tools
• Alerts can route to email, SMS, or a webhook that triggers an automation flow
• Role-based access (Admin / Editor / Viewer) controls who can edit dashboards versus just view them

Grafana sits downstream of ingestion and storage. See our MQTT explainer and IoT stack overview for how readings get to it in the first place, or talk to us about wiring up a dashboard for your own sites.