ERP5 min readNetray Engineering TeamUpdated March 4, 2025

MES-ERP Integration for Discrete Manufacturing: Architecture, Patterns, and Pitfalls

MES-ERP integration is the bidirectional exchange of production orders, material consumption, labor, and quality data between a Manufacturing Execution System and an ERP such as Infor SyteLine, Infor LN, or Infor M3. The ERP owns the plan - orders, BOMs, routings, standards - while the MES owns real-time execution on the shop floor. When the two are integrated well, work-in-process visibility, inventory accuracy, and actual costing improve dramatically; when they are not, planners schedule against stale data and finance closes the month on estimates. This guide covers the integration layers, data mapping, middleware choices, and where AI agents remove the manual glue work.

The ISA-95 Model: What Belongs in ERP vs. MES

ISA-95 (IEC 62264) defines the boundary: Level 4 (ERP) handles order management, master scheduling, standard costing, and procurement; Level 3 (MES) handles dispatching, detailed sequencing, data collection, and genealogy. Discrete manufacturers get into trouble when they blur this line - for example, maintaining routings in both systems or letting the MES hold inventory balances the ERP never sees. The clean pattern is: ERP releases a production order with BOM, routing, and standards; MES executes it and returns operation completions, material issues, labor bookings, and scrap by reason code. Systems like Infor MES (formerly Lighthouse), Siemens Opcenter, Rockwell Plex, and Critical Manufacturing all publish B2MML-based interfaces built on this model, which maps cleanly to SyteLine job orders (job, jobroute, jobmatl tables) and LN production orders (tisfc sessions).

Integration Patterns: Flat Files to Event-Driven APIs

Most brownfield plants still run nightly flat-file exchanges, which means planners work against data up to 24 hours old. Modern integrations are event-driven: the MES posts transactions to the ERP within seconds using REST or messaging. For Infor products, Infor ION is the standard bus - MES events map to ION BODs (Business Object Documents) like ProductionOrder, InventoryCount, and OperationActivity. For SyteLine specifically, IDO-based REST endpoints (LoadCollection and UpdateCollection against SLJobs, SLJobRoutes, SLJobmatls) give sub-second writes without custom stored procedures.

Batch flat-file: lowest cost (roughly $15,000-40,000 to build) but introduces 4-24 hour latency and reconciliation overhead
Middleware/ESB via Infor ION or MuleSoft: $60,000-150,000 typical, near-real-time, with retry queues and BOD-level audit trails
Direct API (IDO REST for SyteLine, LN ION APIs, M3 API/MI programs): fastest, but tightly couples MES release cycles to ERP versions
Event streaming (Kafka or Azure Event Hubs): best for multi-plant rollups above 50,000 transactions per day, adds infrastructure skill requirements

The Five Data Flows That Must Reconcile

Every MES-ERP project succeeds or fails on five transaction flows, and each needs an owner, a reconciliation report, and an error queue. Backflush conflicts are the classic failure: if the ERP backflushes material at operation complete while the MES also reports actual issues, inventory doubles down. Decide per-item whether consumption is MES-reported or ERP-backflushed, and enforce it in the item master (in SyteLine, the Backflush flag on jobmatl; in LN, the phantom and backflush settings on the BOM line).

Order release downstream: production order, BOM effectivity, routing steps, and drawing revision must arrive before the MES dispatches
Material consumption upstream: lot and serial references must match ERP lot masters exactly or genealogy breaks
Labor and machine time upstream: map MES reason codes to ERP indirect codes so variance reporting stays meaningful
Completions and scrap upstream: scrap by reason code drives ERP yield planning and quality cost buckets

Realistic Timelines, Costs, and Failure Modes

A single-plant MES-ERP integration with an established MES typically runs 4-9 months: 6-8 weeks for data mapping and master data cleanup, 8-12 weeks for interface build and unit test, and 6-8 weeks for parallel run and cutover. Budget $120,000-350,000 all-in for a mid-size discrete plant, with 30-40 percent of effort going to master data - part numbers, work centers, and reason codes that do not match between systems. The most common failure mode is skipping the parallel run: plants that cut over cold report 3-6 months of inventory record accuracy below 90 percent. Version lock-in is the second trap - hard-coded stored procedure integrations against SyteLine 9 or LN 10.5 tables routinely break during CloudSuite upgrades, adding $40,000-80,000 of rework that API-based designs avoid.

How Netray AI Agents Accelerate MES-ERP Integration

Netray builds and maintains MES-ERP integrations for SyteLine, LN, and M3 environments using AI agents that do the tedious 40 percent humans get wrong. Our mapping agents read both schemas - MES database or B2MML export and ERP IDOs or table layouts - and generate a field-level mapping workbook with type mismatches and missing master data flagged before build starts, typically compressing the mapping phase from 6 weeks to under 2. Reconciliation agents run continuously after go-live, comparing MES completions to ERP job transactions every 15 minutes and auto-opening exceptions with root-cause context, which clients use to hold inventory record accuracy above 98 percent. Because Netray specializes in aerospace and defense discrete manufacturing, agents also enforce ITAR data-handling boundaries so shop floor data never routes through non-compliant cloud services.

Frequently Asked Questions

What is the difference between MES and ERP in manufacturing?

ERP (Level 4 in ISA-95) manages business planning: orders, purchasing, scheduling, inventory valuation, and costing. MES (Level 3) manages real-time execution: dispatching work to machines, collecting labor and material transactions, tracking WIP, and recording quality results. The ERP tells the plant what to make and at what standard cost; the MES records what actually happened. Integration keeps both aligned so planning runs on real data and finance closes on actuals.

How long does an MES-ERP integration project take?

A single-plant integration with an established MES and ERP typically takes 4-9 months end to end: 6-8 weeks of data mapping and master data cleanup, 8-12 weeks of interface development and testing, and 6-8 weeks of parallel run before cutover. Multi-plant rollouts add 2-4 months per additional site. Projects that skip master data cleanup or the parallel run routinely double their timeline in post-go-live firefighting.

Does Infor SyteLine have a built-in MES?

Infor offers Infor MES (the former Lighthouse Systems product) as a companion to SyteLine and CloudSuite Industrial, with pre-built ION integration for production orders, labor, and material transactions. SyteLine itself includes basic shop floor functions - job transactions, labor reporting, and the Factory Track module for barcode data collection - which are sufficient for simpler plants but lack full MES capabilities like detailed dispatching, machine connectivity, and electronic work instructions.

Key Takeaways

1The ISA-95 Model: What Belongs in ERP vs. MES: ISA-95 (IEC 62264) defines the boundary: Level 4 (ERP) handles order management, master scheduling, standard costing, and procurement; Level 3 (MES) handles dispatching, detailed sequencing, data collection, and genealogy. Discrete manufacturers get into trouble when they blur this line - for example, maintaining routings in both systems or letting the MES hold inventory balances the ERP never sees.
2Integration Patterns: Flat Files to Event-Driven APIs: Most brownfield plants still run nightly flat-file exchanges, which means planners work against data up to 24 hours old. Modern integrations are event-driven: the MES posts transactions to the ERP within seconds using REST or messaging.
3The Five Data Flows That Must Reconcile: Every MES-ERP project succeeds or fails on five transaction flows, and each needs an owner, a reconciliation report, and an error queue. Backflush conflicts are the classic failure: if the ERP backflushes material at operation complete while the MES also reports actual issues, inventory doubles down.

Talk to Netray about an MES-ERP integration assessment - we will map your data flows and flag the reconciliation gaps in two weeks.

Talk to an Expert See a Demo

Related Resources

ERP