What are the four types of integration?

There is no single universal standard for “four types of integration.” Different textbooks and vendors use the phrase to mean different things (for example: vertical vs horizontal, internal vs external, etc.). In industrial and regulated environments, the most practical way to think about four integration types is along these dimensions:

1. Data integration

Data integration focuses on moving and harmonizing data between systems so it can be used consistently.

• Scope: Master data, transactional data, equipment data, quality records, and historical time-series data.
• Typical examples: Moving production results from MES to ERP; pulling equipment tags from a historian into an analytics platform; synchronizing part numbers and BOM identifiers across PLM, ERP, and MES.
• Common mechanisms: Batch ETL, streaming pipelines, APIs, database replication, flat-file interfaces.
• Key constraints in regulated plants: Data integrity rules, audit trails, versioning of schemas and mappings, validated reports, and long-term retention requirements.

Failure modes include silent mapping errors, duplicate or missing records, loss of context (e.g., losing linkages between lot, serial, and process), and broken downstream reports. These typically show up late, which is why test coverage, traceability of transformations, and controlled migration plans are critical.

2. Process and workflow integration

Process integration connects business and shop-floor workflows across systems, so that a multi-step process functions coherently end-to-end.

• Scope: Order-to-manufacture, engineering change, nonconformance and CAPA handling, maintenance work order cycles, supplier approvals.
• Typical examples: Automatically creating a production order in MES when an ERP order is released; triggering a quality workflow when a test result fails; updating maintenance status in EAM/CMMS based on machine events.
• Common mechanisms: Workflow engines, BPM tools, orchestration layers, event-driven integrations, message queues.
• Key constraints in regulated plants: Documented procedures, e-signature rules, segregation of duties, and the need to prove that workflows behave consistently after changes.

Failure modes include broken handoffs between systems, orphaned work items, conflicting process versions across sites, and workarounds outside the system (spreadsheets, email). These often undermine compliance, traceability, and metrics. Any change here usually requires impact assessment, SOP updates, and re-training.

3. Application integration

Application integration handles how entire software applications interoperate while each remains a distinct system of record.

• Scope: ERP, MES, PLM, QMS, LIMS, WMS, EAM/CMMS, data historians, and analytics tools.
• Typical examples: ERP–MES integration for orders, materials, and confirmations; PLM–MES integration for routing and work instructions; QMS–MES integration for nonconformance data and CAPA triggers; LIMS–MES integration for sample requests and results.
• Common mechanisms: REST/SOAP APIs, message buses, integration platforms (iPaaS), vendor-specific connectors, and occasionally point-to-point flat-file exchanges.
• Key constraints in regulated plants: Validated systems, vendor qualification, change control across multiple owners, and multi-decade application lifecycles.

Failure modes include tight point-to-point couplings that make upgrades risky, integration logic buried in custom code with poor documentation, and inconsistent master data definitions between applications. Full replacement of a major application purely to “simplify integration” often fails in heavily regulated environments because of revalidation cost, downtime, and the need to re-establish all historical traceability.

4. Physical / OT (operational technology) integration

Physical or OT integration links the shop floor and test equipment to higher-level systems.

• Scope: PLCs, CNCs, test stands, robots, sensors, HMIs, data acquisition systems, and industrial networks.
• Typical examples: Reading machine states and counters into MES; sending recipes or NC programs from MES/PLM to equipment; collecting detailed process parameters in a historian; connecting vision systems for automated inspection.
• Common mechanisms: Industrial protocols (OPC UA, Modbus, proprietary drivers), edge gateways, historians, and vendor-specific middleware.
• Key constraints in regulated plants: Long equipment lifecycles, vendor lock-in, limited ability to modify validated equipment, cybersecurity controls, and very limited downtime windows.

Failure modes include unstable drivers, protocol mismatches after firmware upgrades, bottlenecks at a single integration gateway, and changes to equipment behavior that unintentionally affect validated processes. These issues often cannot be fixed quickly due to qualification and safety considerations, so designs should assume coexistence with legacy controls and gradual evolution.

Why this framing matters in brownfield, regulated environments

Most plants operate with a mix of old and new systems across IT and OT. In practice, any integration initiative cuts across all four types:

• Adding a new MES impacts application integration and usually data and process integration.
• Pulling data from legacy equipment introduces OT integration and often requires intermediate historians or gateways.
• Automating quality workflows touches process integration and must respect QMS constraints and validation.

Attempting to solve integration problems by fully replacing legacy systems is high risk. In aerospace-grade or similar environments, requalifying new systems, migrating historical data, revalidating reports, and coordinating downtime typically exceed expectations in cost and schedule. Incremental, well-scoped integration across these four types tends to be more realistic.

Other “four types of integration” you might see

In some materials you may encounter other groupings, such as:

• Vertical, horizontal, internal, external integration.
• Data, functional, business, organizational integration.

These can be useful for high-level discussion, but for planning real-world projects in a regulated, brownfield environment, explicitly separating data, process, application, and OT integration makes dependencies, risks, and ownership clearer.