BSBurak Soykan
Industrial IoT7 min read

Designing an industrial telemetry pipeline from machine data to MES-ready payloads

A practical structure for turning raw machine, database or controller data into stable telemetry that can feed dashboards, OEE screens, MES integrations and runtime monitoring flows.

Problem

Industrial data is rarely clean at the source. A machine may expose counters, states, alarms or program data through a controller protocol, a local database, a vendor SDK or a gateway service. The first engineering task is not visualization; it is building a reliable acquisition and normalization path.

Pipeline model

  1. Acquire: read from CNC, PLC, database, API or local service.
  2. Normalize: map vendor-specific values into stable operational states.
  3. Shape: create payloads with machine identity, timestamp, status, counters and alarm context.
  4. Transport: publish through MQTT, REST, WebSocket or MTConnect depending on the target environment.
  5. Operate: add logs, health checks, restart policy, configuration files and deployment documentation.

Payload design

{
  "machineId": "CNC-01",
  "state": "START",
  "program": "O1234",
  "counter": 12840,
  "alarm": null,
  "timestamp": "2026-05-18T10:25:00Z"
}

Validation

A telemetry pipeline is useful only if it can be operated after deployment. The validation checklist should include reconnect behavior, duplicate handling, timestamp consistency, offline state mapping, log rotation, credential protection and service recovery.

Back to Blog