In most plants, data still travels point to point: PLC to SCADA, SCADA to MES, MES to ERP, each link with its own format and engineering. A Unified Namespace (UNS) changes the route: every system publishes to one shared, named structure and reads what it needs from there. This article explains, from vendors' public documentation and the Eclipse Sparkplug 3.0 specification, what a UNS is made of, how
1. What a Unified Namespace is
A Unified Namespace is a standard way to organise and name the data of a plant or a whole enterprise, so that its structure and its events sit behind one communication interface. That is how Inductive Automation defines it; its article credits the term to Walker Reynolds, who describes it as a real-time single source of truth, organised the way the business is organised and not tied to any one product.
The reason it exists is the point-to-point chain. Inductive Automation describes the usual layered architecture: plant floor to SCADA, SCADA to MES, MES to ERP, often in different data formats, each layer needing specialised engineering to talk to the next. In a UNS, data goes through the hub instead of layer by layer, and producers and consumers are added by connecting them to the same infrastructure. Older devices built for point-to-point links do not have to be replaced first: an IIoT gateway can publish to the UNS on their behalf.
2. A design pattern, not a product
No single program is "the UNS". Inductive Automation lists the parts a UNS combines: a software platform that manages it, an edge platform with device and network connections at the OT level, connections to sensors, PLCs and business applications, an MQTT broker that moves data between them, and a central SQL database or cloud data service. Sepasoft describes the same idea from the MES side: a central structure where systems publish and subscribe in real time, usually on MQTT and Sparkplug, organised by a shared topic hierarchy, so that systems are not mapped to each other one pair at a time.
In practice a UNS is three decisions: a naming model, a publish/subscribe transport, and rules on who publishes what. Software choices follow.
3. The naming model: ISA-95 as the tree
Inductive Automation's webinar slides on building a UNS with Ignition give three tips: model the architecture on your own facility and processes, follow what people in operations want to see, and base the naming conventions on your hierarchy. Its example structure is the
Plant1/Packing/Line2/Filler1/StatePlant1/Packing/Line2/Filler1/GoodCountPlant1/Utilities/Compressor3/DischargePressure
The data model sits under those names. In Ignition it is built with user-defined types (UDTs): every instance inherits a change to the definition, and nested UDTs form directories that map onto the MQTT topic structure. N3uron uses templates the same way, with nesting, and custom properties set where each instance reads its data.
One detail matters when Sparkplug is used. MQTT itself does not define a topic namespace, but Sparkplug does, and it is fixed: spBv1.0/group_id/message_type/edge_node_id/[device_id]. The Group ID groups edge nodes, and the specification gives a cell or a production line as typical plant-floor groupings. Deeper levels go into the metric name, which may be hierarchical, such as Folder 1/Folder 2/Metric Name, so that consuming applications can rebuild the folders. With plain MQTT and JSON payloads, the topic path itself can carry the tree. Decide which before naming anything.
4. MQTT and what Sparkplug adds
MQTT is publish/subscribe. A client connects to the MQTT server (the broker) without knowing the other clients, subscribes to what it wants and publishes what it has. The Sparkplug specification calls this decoupling of devices from any one consumer application a principal idea of IIoT. MQTT leaves the topic names and the payload format open; Sparkplug defines both for SCADA and IIoT use:
- Topic namespace and message types. Births, deaths, data and commands for edge nodes and devices (NBIRTH, NDEATH, DBIRTH, DDEATH, NDATA, DDATA, NCMD, DCMD) and STATE for host applications. Each Group ID and Edge Node ID pair must be unique on the broker.
- State awareness. A birth certificate is the first message an edge node publishes and carries all current values. Its death certificate is registered with the broker as the MQTT Will message, so subscribers receive it even when the connection drops without warning.
- Report by exception. After the birth messages, values are published only when they change. The specification says edge nodes should not publish periodically as a general rule.
- Payload. Sparkplug B encodes metrics in binary with Google Protocol Buffers, with timestamps, data types and metadata.
The specification's own architecture figure puts the primary host application (often called the SCADA host), the historian, MES and analytics on the IT side as Sparkplug clients, and the edge nodes with their devices on the OT side. How a primary host and store and forward work together between N3uron and Canary is covered in N3uron to Canary Historian over MQTT Sparkplug B.
5. The broker holds the present, not the past
Inductive Automation is direct: an MQTT broker should not be used as the UNS store, because brokers are not designed to keep data long term and doing so makes the broker slow. Its solution page puts it the other way round: the broker holds the real-time data, and a historian adds the history.
So a process historian sits behind the broker as one more subscriber. Canary's product page presents Virtual Views as its UNS and asset definition layer, and Canary's architecture article explains that Virtual Views apply structure changes, tag aliasing and asset models on top of the Historian without altering the stored archive; see Canary Historian architecture, and for how queries shape the answer, historian retrieval modes.
6. Production context: Sepasoft's argument
Sepasoft makes a case worth knowing before designing a UNS; it is Sepasoft's view, not a rule. Its blog argues that most UNS implementations carry machine-layer data: equipment status, sensor readings, setpoints and control outputs. That answers whether a machine is running, but not what product was running when it faulted, or whether a temperature was within the recipe's tolerance. A bill of materials, or a batch that references several lots each with its own genealogy, does not fit a simple tag and value.
According to Sepasoft, that context lives in MES data: batch recipes and production orders, quality results and electronic batch records, downtime reasons and operator notes, maintenance and lot genealogy. It proposes preparing MES data in a structured, contextualised form before it is published, calls the result an Extended Unified Namespace, and names its SepaIQ platform for that job. Where the orders and lots come from in the first place is the level 3 / level 4 boundary of
7. What each platform documents
Only what each vendor's documents say, without a ranking. Ignition can host the MQTT broker; its manual names a third-party MQTT server as the alternative.
| Platform | What its documents describe for a UNS |
|---|---|
| Ignition (8.3 manual, IIoT Architecture; UNS solution page) | Publishing from the edge with MQTT Transmission, a broker with MQTT Distributor or a third-party server, subscribing with MQTT Engine; UDTs as the data model; Historian Core and Perspective or Vision in the typical module set. The MQTT modules come from Cirrus Link |
| N3uron (KB 1.22) | An edge node: Sparkplug Client publishes any tag, template instances included, as Sparkplug B v2 or v3, by exception, with store and forward, several brokers and outbound TLS; MQTT Client for other payloads; templates with inheritance and nesting |
| Canary | Historian behind the broker; collectors for OPC, MQTT and SQL; Virtual Views as UNS and asset definition over stored tags; real-time feeds out over MQTT Sparkplug |
| Sepasoft | Production context from MES: orders, recipes, quality, downtime, genealogy; SepaIQ as a data preparation layer, as Sepasoft describes it |
8. Security
Inductive Automation notes that a UNS connects devices to the whole enterprise, so security matters more. It describes a demilitarized zone between the business and plant networks, a buffer between two firewalls where the MQTT broker and its connections to the edge live, plus TLS on MQTT, which needs a certificate from a certificate authority. The broker authenticates each client and then checks an access control list of which topics it may read or write.
The Sparkplug specification's security chapter (guidance only) adds detail: each client has a unique Client ID and optional user name and password, sockets should be encrypted with TLS when data crosses public networks, and broker ACLs can restrict each edge node to its own topics, for example publishing only its own birth, data and death messages, or not subscribing to device commands at all when devices must not be written. Default ports in N3uron's documentation are 1883 for MQTT and 8883 for MQTTS. For zones and conduits see IEC 62443; for Rockwell systems, FactoryTalk and Logix firewall ports.
What we check first
This is TDL's practice, not a requirement of any vendor. Before choosing software for a UNS, we check:
- The naming of one line's assets and tags, agreed with operations, before anything is published.
- Which PLCs and protocols are already in the plant and which driver reads each; see PLC drivers on Ignition, Optix, AVEVA, N3uron and Kepware, N3uron with ControlLogix and CompactLogix and the Ignition Logix driver.
- Which data needs history and which only its current state.
- Who owns the broker, who patches it and who issues its certificates.
- Which network zone the broker sits in, and which connections cross each firewall.
- Where production context comes from: orders, batches and lots, and which system holds the master copy.
How TDL approaches it
TDL Industrial Automation is an Ignition Certified Integrator, an N3uron Certified Integrator, a Canary Certified Partner and a Sepasoft MES Certified Integrator. When we plan a UNS, we start from the plant's existing PLCs and network, and decide where the edge, broker and historian sit before choosing software. More on these services: Ignition, N3uron, Canary Historian and MES on Ignition with Sepasoft.
Frequently asked questions
Is a Unified Namespace the same thing as an MQTT broker?
No. The broker moves messages between publishers and subscribers. Inductive Automation advises against using a broker as the UNS store because brokers are not designed to keep data long term; the UNS is the naming model and the data organised under it, with a historian or database keeping the history.
Does a UNS have to use Sparkplug B?
No. Inductive Automation calls MQTT with Sparkplug the most common UNS architecture, not the only one. Sparkplug fixes the topic structure and payload; with plain MQTT and JSON, the topic path can carry the plant hierarchy, but MQTT itself defines no payload format, so every consumer must agree on one.
How does ISA-95 fit into a Unified Namespace?
Where is history kept in a UNS?
In a historian or database that subscribes to the broker. Inductive Automation's documents describe the broker as holding real-time data and a historian adding historical context; Canary describes Virtual Views as a UNS and asset model built over the tags it has stored.
Does a UNS replace SCADA or MES?
No. In the sources used here, SCADA, the historian, MES and ERP are publishers and consumers of the UNS. Sepasoft argues that MES is where the production context comes from: orders, recipes, quality, downtime and genealogy.
Sources
- Inductive Automation, UNS: Unified Namespace, article, last updated 7 July 2026
- Inductive Automation, Ignition Unified Namespace solution, solution page
- Inductive Automation, Ditch Data Silos: Create a Unified Namespace With Ignition UDTs & MQTT, webinar page (last updated 14 January 2026) and its slide PDF
- Inductive Automation, IIoT Architecture and User Defined Types - UDTs, Ignition 8.3 user manual
- Eclipse Foundation, Sparkplug Specification, version 3.0.0, 16 November 2022 — sections 1.1, 2.1–2.4, 3, 4.1, 6.2, 6.4.2, 7
- N3uron, Sparkplug Client: Introduction and Configuration, N3uron Knowledge Base 1.22
- N3uron, MQTT Client: General features, N3uron Knowledge Base 1.22
- N3uron, Templates Overview, Templates Inheritance and Templates Nesting, N3uron Knowledge Base 1.22
- Canary Labs, Canary System product page
- Canary Labs, Understanding Canary System Architecture Options (version 25) and Data Contextualization, Canary Community knowledge base
- Sepasoft, Bethany Duff, Why Your Unified Namespace Needs MES Context, Sepasoft blog, page read 22 November 2026
Read 22 November 2026. This article paraphrases the sources and does not quote them; read the original documents before designing or working on a system.