Telemetry to alert
This workflow is the platform’s hot path: millions of devices publish readings, and within seconds those readings must appear on dashboards and, when a rule trips, turn into an alert in an operator’s inbox.
Latency budget
End-to-end target is < 5 seconds from device publish to dashboard update, and < 10 seconds from a threshold breach to a delivered alert.
Participants
MQTT Broker
Accepts device sessions and fans telemetry to consumers.
Stream Processor
Windows and aggregates the firehose into metrics.
Rule Runtime
Evaluates thresholds and anomaly signals.
Steps
Publish
The Edge Agent batches readings and publishes to the broker.
Aggregate
The Stream Processor consumes the topic and computes windowed aggregates.
Evaluate
Enriched events are passed to the rules, which decide whether a threshold or anomaly condition is met.
Notify & render
A live aggregate is pushed to the dashboard; a breach additionally raises an alert via the notification provider.
The same aggregate that updates charts also feeds rule evaluation, so what an operator sees and what triggers an alert are always derived from one source.