Data Architecture
Unified Namespace and Industrial Ontologies
A unified namespace gives factory systems a shared, real-time address structure. An industrial ontology defines what the entities mean and how assets, orders, materials, events and people relate. Together they turn disconnected tags into operational context.
Enterprise
Company context
Site
Factory context
Area · line
Process context
Asset · signal
Operational event
What is a unified namespace in manufacturing?
A unified namespace, usually shortened to UNS, is a shared and standardized structure for operational events. Producers publish data to agreed paths. Consumers subscribe to the data they need. The namespace gives SCADA, MES, ERP, analytics and AI the same live view of the factory.
A common hierarchy follows the ISA-95 pattern of enterprise, site, area, line, asset and signal. The hierarchy is a starting point, not a substitute for governance. Names, schemas, units, ownership, retention and permissions still need explicit rules.
What a unified namespace is not
- It is not a replacement for ERP, MES, SCADA, a historian or a data warehouse.
- It is not a license to publish every raw PLC tag without context.
- It is not automatically a single source of historical truth.
- It is not complete governance unless schemas, access and ownership are enforced.
The UNS is most useful as a current-state and event distribution layer. Historical stores, transactional systems and analytical platforms continue to serve different purposes.
What is an industrial ontology?
An industrial ontology is a machine-readable model of concepts and relationships in an operation. It can define that a press is an asset, a press belongs to a line, a line executes an operation, an operation consumes a material lot, and a quality result applies to the produced lot.
- Entity
- A thing with an identity, such as an asset, order, product, tool or person.
- Property
- A value that describes an entity, such as nominal speed or material grade.
- Relationship
- A typed connection between entities, such as produces, belongs to or depends on.
- Constraint
- A rule that limits valid structures or values.
How do OPC UA information models fit?
OPC UA provides an industrial communication infrastructure and an object-based information model. Companion Specifications define reusable domain models for machines, devices and industry-specific processes. This supports semantic interoperability: two systems can exchange data with a shared understanding of the object types, variables, alarms and relationships involved.
A factory ontology can reuse these standard models and extend them with company-specific concepts. The aim is not to model every possible fact. It is to model the relationships needed by real analytical and operational decisions.
When do you need a UNS, an ontology, or both?
| Layer | Primary role | Best suited to |
|---|---|---|
| Unified namespace | Distribute current state and events | Low-latency integration and shared operational topics. |
| Ontology or knowledge graph | Define meaning and relationships | Cross-system queries, reasoning, lineage and AI grounding. |
| Historian or time-series store | Retain measurements over time | Trends, replay, diagnostics and model training. |
| Warehouse or lakehouse | Combine analytical datasets | Large-scale reporting, feature engineering and cross-plant analysis. |
How should the model be governed?
- Assign an owner to every namespace domain and reusable model.
- Version schemas and preserve backward compatibility where possible.
- Validate data types, units, required fields and allowed ranges at the source boundary.
- Separate raw signals from approved, contextualized data products.
- Document external references to ERP, MES and maintenance identifiers.
- Test the model with real queries before expanding its scope.
Frequently asked questions
Is MQTT required for a unified namespace?
MQTT is a common transport because its publish-subscribe model fits event distribution, but the architectural principle is broader. Governance, shared naming and context matter more than a single protocol.
Is an ontology the same as a data model?
An ontology is a type of semantic data model focused on concepts and relationships. A database schema may describe storage without expressing the same operational meaning.
Should every PLC tag enter the namespace?
No. Publish data that has an identified consumer, owner and meaning. Unfiltered tag replication increases cost and ambiguity without improving decisions.
Continue learning