IoT-to-ERP Integration for Manufacturing Operations
IoT-to-ERP integration connects machine controllers, sensors, and shop floor devices to enterprise transactions, so production counts, downtime events, and quality readings post to ERP work orders without an operator keying them. The architecture that survives contact with a real plant is layered: devices speak OPC UA, MTConnect, or Modbus to an edge gateway, the gateway normalizes and buffers, a broker such as MQTT distributes, and only aggregated, business-meaningful events reach the ERP. Sending raw telemetry into ERP is the mistake that sinks most first attempts.
The Layered IoT-to-ERP Architecture
Four layers, each with a distinct job. Devices and PLCs expose data over OPC UA, MTConnect for machine tools, or Modbus TCP for older equipment. An edge gateway - Kepware, Ignition, or a Linux gateway running an OPC UA client - polls, normalizes tag names, and buffers locally so a network outage does not lose a shift of production. A broker or streaming layer, typically MQTT with Sparkplug B, distributes to subscribers. Finally an integration service aggregates telemetry into business events and posts them to ERP through supported APIs. The ERP sees a completed quantity of 240 for a work order operation, not 240 individual cycle-complete signals.
What Shop Floor Data Belongs in the ERP
The filter is simple: if it changes an ERP record that someone plans, costs, or ships against, it belongs. If it only informs an engineer looking at a trend, it belongs in a historian or time-series database instead. Manufacturers who ignore this end up with ERP transaction tables growing by hundreds of millions of rows a year and no analytical benefit to show for it.
- Production counts posted as work order operation completions, aggregated per shift or per hour
- Downtime events with a mapped reason code, posted against the work center for OEE and capacity
- In-process quality readings that trigger an ERP nonconformance or inspection record
- Consumption signals for backflush confirmation where material issue is machine-driven
Protocols and Gateways: OPC UA, MQTT, and MTConnect
Protocol choice is dictated by the equipment you already own, not by architectural preference. OPC UA is the general standard for industrial equipment and provides typed data models plus security; MTConnect is the norm for CNC machine tools and is read-only, which plant engineers appreciate; Modbus TCP covers legacy equipment with no semantics at all, so all meaning lives in your tag mapping. MQTT is the transport of choice above the gateway because it is lightweight, supports quality-of-service levels, and handles intermittent connectivity well.
- OPC UA for most modern PLCs and equipment; use certificate-based security, not anonymous sessions
- MTConnect for CNC machine tools - read-only by design, which simplifies plant IT approval
- Modbus TCP for legacy assets; document every register mapping because the protocol carries no metadata
- MQTT with Sparkplug B above the gateway for birth/death certificates and stateful topic structure
Latency, Volume, and Cost Control
A single CNC machine publishing at one hertz produces roughly 86,400 messages per day, and a hundred machines produce 8.6 million. None of that should reach the ERP. Aggregate at the edge or in the streaming layer, and define an explicit posting cadence per event type: production counts hourly or at operation completion, downtime on event with a minimum duration threshold of two to five minutes to filter micro-stops, quality on exception. Buffer at the edge for at least 24 hours so a plant network outage or an ERP maintenance window causes delay rather than data loss. Keep raw telemetry in a time-series store where retention is cheap and query patterns actually fit.
How Netray Builds IoT-to-ERP Pipelines
Netray designs the aggregation boundary first, because that decision determines cost and ERP health for years. We build edge normalization and buffering, define the business event catalog, and implement posting into Infor SyteLine, LN, or M3 through supported APIs with idempotent keys so a replay after an outage cannot double-count production. Our AI agents learn each machine's normal signal pattern and flag sensors that have silently drifted or stopped reporting - a failure mode that otherwise shows up weeks later as quietly wrong OEE numbers. Clients typically eliminate manual production reporting at the terminal and gain downtime reason data accurate enough to act on.
Frequently Asked Questions
How do you connect shop floor machines to an ERP system?
Use a layered architecture rather than a direct connection. Machines expose data through OPC UA, MTConnect, or Modbus to an edge gateway that normalizes tags and buffers locally. An MQTT broker distributes the data, and an integration service aggregates it into business events - operation completions, downtime records, quality exceptions - which post to ERP through supported APIs. Raw telemetry should never reach the ERP directly.
What is the difference between OPC UA and MQTT?
OPC UA is an industrial protocol for accessing device data with typed information models and built-in security; it is how you talk to PLCs and equipment. MQTT is a lightweight publish-subscribe transport for moving messages between systems, well suited to unreliable networks. They are complementary: an edge gateway typically reads equipment over OPC UA and publishes normalized data over MQTT to subscribers upstream.
Should IoT data be stored in the ERP database?
No. Store raw telemetry in a time-series database or historian where retention is inexpensive and query patterns match the data. Post only aggregated, business-meaningful events to ERP - operation completions, downtime with reason codes, quality exceptions. A hundred machines publishing once per second generate millions of daily messages, and pushing that into ERP transaction tables degrades performance without delivering analytical value.
Key Takeaways
- 1The Layered IoT-to-ERP Architecture: Four layers, each with a distinct job. Devices and PLCs expose data over OPC UA, MTConnect for machine tools, or Modbus TCP for older equipment.
- 2What Shop Floor Data Belongs in the ERP: The filter is simple: if it changes an ERP record that someone plans, costs, or ships against, it belongs. If it only informs an engineer looking at a trend, it belongs in a historian or time-series database instead.
- 3Protocols and Gateways: OPC UA, MQTT, and MTConnect: Protocol choice is dictated by the equipment you already own, not by architectural preference. OPC UA is the general standard for industrial equipment and provides typed data models plus security; MTConnect is the norm for CNC machine tools and is read-only, which plant engineers appreciate; Modbus TCP covers legacy equipment with no semantics at all, so all meaning lives in your tag mapping.
Put this into numbers
Free interactive tools for exactly this problem. No signup to use them.
Shop Floor Digitization Scorecard
A 10-question scorecard measuring how much of your shop floor still runs on paper - and which digitization gaps are costing you the most.
Free ToolManufacturing Downtime Cost Calculator
Combine lost contribution margin, idle labor, and absorbed overhead into a defensible monthly and annual downtime cost - with recovery effects modeled.
Free ToolCycle Time & Capacity Calculator
Translate cycle time, parallel stations, shift pattern, and realistic utilization into hourly, daily, and annual production capacity.
Terms used in this article
Want machine data posting straight to ERP work orders? Netray can design the edge-to-ERP pipeline and the aggregation rules that keep it affordable.
Related Resources
An API-First ERP Integration Strategy
An API-first ERP integration strategy replaces nightly file drops with REST and event APIs. Learn contract design, versioning, throttling, and security.
ERPIntegration Error Handling Patterns
Integration error handling patterns for ERP: retries with backoff, idempotency keys, dead-letter queues, poison-message handling, and replay you can trust.
ERPERP-WMS Integration Guide
ERP-WMS integration guide: design the order, receipt, and inventory interfaces between your ERP and warehouse system, and keep stock balances reconciled daily.