Every greenfield factory faces the same data integration nightmare — just faster. The MES needs data from 40 PLCs. The ERP needs production counts from the MES. The AI models need sensor streams from every asset. The CMMS needs vibration data to predict failures. The OEE dashboard needs availability signals from SCADA. The energy system needs load data from every motor. And every single one of these connections is built as a custom, point-to-point integration — a spaghetti architecture where adding one new system means building N new connections to every existing system. This is the architecture that 80% of factories still run. And it's the reason why adding AI to a brownfield plant takes 12-18 months and costs $2-8M in integration alone. The Unified Namespace (UNS) eliminates this entirely. It creates one event-driven data bus where every system publishes once and any consumer subscribes — no point-to-point integrations, no data silos, no vendor-specific middleware. For greenfield factories, designing UNS from day one is the single highest-leverage architecture decision you will make. It's the foundation that every AI model, every digital twin, every agentic system depends on. Without it, your smart factory is just an expensive collection of disconnected smart devices. Book a 30-minute UNS architecture session to see how this applies to your greenfield project.
10 systems = 45 custom integrations. 20 systems = 190 integrations. Adding one new AI model means building connections to every data source individually. Each integration uses a different protocol, format, and middleware. Every vendor upgrade breaks something.
10 systems = 10 connections (each to the broker). 20 systems = 20 connections. Adding one new AI model means one subscription to existing topics. Data arrives in standardized format. Vendor swaps require zero re-integration.
What a Unified Namespace Actually Is
A UNS is a single, event-driven data bus — organized as a hierarchical topic tree — that serves as the single source of truth for all operational data in your factory. Producers (PLCs, sensors, machines) publish data to topics. Consumers (MES, ERP, AI, CMMS, dashboards) subscribe to the topics they need. The broker ensures every message reaches every subscriber in real-time. No polling. No batch transfers. No point-to-point connections.
Full topic path: AcmeCorp/Chicago-Plant-1/Assembly/Line-04/Robot-Cell-A/temperature/bearing-1
The topic tree mirrors your physical asset hierarchy — typically following ISA-95 structure: Enterprise → Site → Area → Line → Cell → Equipment → Tags. This hierarchy provides context: when the AI model reads a temperature value, it knows exactly which bearing, on which robot, on which line, in which plant the reading comes from. Context travels with the data. No manual mapping. No lookup tables.
Want to see how your factory's asset hierarchy maps to a UNS topic tree? Book a free UNS design session — we'll build the first draft of your topic structure in 30 minutes.
The Technology Stack: MQTT + OPC UA + Sparkplug B
UNS is an architectural pattern, not a product. It's implemented using open protocols that work together at different layers:
OPC UA
Collects data from PLCs, SCADA, and machine controllers with rich semantic modeling. Provides the "dictionary" — data types, units, relationships between variables. Runs locally within each automation cell.
OPC UA → MQTT Translation
Edge gateways normalize OPC UA data into MQTT messages with Sparkplug B payload structure. This is the bridge between the machine-level protocol and the enterprise-level transport. Runs on edge computing infrastructure near the production floor.
MQTT Broker (The UNS)
The heart of the architecture. All producers publish to the broker. All consumers subscribe from the broker. The broker maintains the topic tree, routes messages, and ensures delivery. Sparkplug B adds state management (birth/death certificates), structured payloads, and auto-discovery. This is your single source of truth.
MES, ERP, AI, CMMS, Dashboards, Digital Twin
Every enterprise system subscribes to the topics it needs. Adding a new consumer (e.g., a new AI model) requires one subscription — not a new integration project. Data arrives in real-time, in a standardized format, with full context.
What UNS Connects: The Full Integration Map
| System | Role in UNS | Publishes | Subscribes To | Without UNS |
|---|---|---|---|---|
| PLCs / Sensors | Producer | Machine state, cycle times, process parameters, sensor readings | Setpoint commands (closed-loop) | Locked in vendor SCADA; data trapped |
| SCADA | Producer + Consumer | Aggregated zone data, alarm events | PLC data for visualization | Point-to-point to each PLC |
| MES | Producer + Consumer | Production orders, batch records, quality results | Machine state, cycle times, counts | Custom API to each data source |
| ERP | Consumer (mostly) | Planning schedules, material availability | Production counts, quality data, energy | Batch file transfers; hours/days lag |
| AI / ML Models | Consumer + Producer | Predictions, anomaly alerts, recommendations | All sensor streams, production data | 12-18 months to integrate each model |
| CMMS | Consumer + Producer | Work orders, maintenance status | Vibration, temperature, runtime data | Manual entry or custom sensor integration |
| Digital Twin | Consumer | Simulation results, optimization recommendations | Real-time state of all physical assets | Partial sync; delayed data |
| OEE Dashboards | Consumer | — | Availability, performance, quality metrics | Manual calculation from multiple sources |
| Energy Management | Consumer | Demand forecasts | Power consumption, utility data, production schedule | Separate metering system; no AI integration |
The column "Without UNS" is what every brownfield factory deals with today. It's why adding AI to an existing plant takes 12-18 months. It's why MES implementations run over budget. It's why OEE dashboards show different numbers than the ERP. UNS eliminates all of these integration problems at the architecture level — but only if it's designed before the systems are installed.
See the full integration map applied to your specific systems. Schedule a UNS architecture demo — we'll map your PLCs, MES, ERP, CMMS, and AI stack into a unified topic tree with real-time data flow.
Why Greenfield Factories Have a Massive UNS Advantage
Implementation Roadmap: Greenfield UNS in 5 Steps
Define the Topic Tree
Factory Design (Step 3)Map your physical asset hierarchy to an ISA-95 topic structure: Enterprise → Site → Area → Line → Cell → Equipment → Tags. Define naming conventions. Document every topic with its data type, unit, update frequency, and owner. This is your UNS blueprint.
Select & Deploy the MQTT Broker
Construction (Step 5)Choose an enterprise-grade MQTT broker with high availability, clustering, and Sparkplug B support. Deploy on edge infrastructure near the production floor. Configure security (TLS, client certificates, ACLs). Size for your device count and message throughput.
Connect Producers via Edge Gateways
Equipment Installation (Step 6)Deploy edge gateways that translate OPC UA from PLCs into MQTT Sparkplug B messages. Configure data mapping: which OPC UA nodes map to which MQTT topics. Enable report-by-exception so only changed values are transmitted — saving bandwidth and compute.
Connect Consumers
Commissioning (Steps 7-8)Subscribe MES, ERP, CMMS, OEE dashboards, AI models, digital twin, and historian to their relevant topics. Each consumer connects once to the broker. Validate end-to-end data flow: PLC → gateway → broker → consumer. Verify data arrives with correct context, format, and timing.
Validate & Evolve
Production (Step 10+)Monitor broker health, message throughput, and consumer lag. Add new topics as new equipment comes online. Expand the namespace to include energy, quality, and logistics data. Retrain AI models on the growing data corpus. The UNS grows with your factory.
Your Factory's Intelligence Starts with the Data Layer
iFactory designs UNS architecture for greenfield factories — from topic tree definition through broker deployment, producer/consumer integration, and ongoing evolution. One data bus. Every system connected. AI-ready from day one.
Frequently Asked Questions
Stop Building Point-to-Point. Start Building UNS.
Every integration you build as point-to-point is technical debt you'll pay for decades. UNS eliminates it from day one. Book a demo to see the architecture that makes AI, digital twins, and agentic systems actually work.
