Unified Namespace

what is a unified namespace (uns)? audience this section is intended mostly for non technical readers who want to understand how plant data flows between systems in mantsu no integration background is required; the concepts are explained from first principles a unified namespace (uns) is an architectural pattern for industrial data integration it defines a single, centralised logical space where every data producer in a plant (sensors, machines, controllers, business systems) publishes its data, and every consumer reads from that same space there is no point to point wiring between systems all data flows through one shared namespace (most commonly a shared centralised message broker) topics in the uns follow a hierarchical naming convention that reflects the physical and logical structure of the plant, commonly based on the isa 95 enterprise hierarchy this naming structure makes data self describing and discoverable any consumer can browse the namespace to understand what data exists, where it comes from, and what it represents, without needing to consult external documentation or negotiate a private interface with the producer the uns acts as the single source of truth for the current and historical state of the plant why use a uns? traditional manufacturing it landscapes grow into a tightly coupled mesh of point to point integrations each system talks directly to the systems it needs scada reads from plcs, mes reads from scada, erp reads from mes, reporting reads from everything every new system added to the plant requires new integrations with every existing system it needs to interact with the result is an integration architecture that is fragile, expensive to change, and nearly impossible to scale the uns pattern solves this by inverting the relationship systems do not talk to each other; they all talk to the centralized namespace problem (point to point) solution (uns) n systems → n² potential integrations n systems → n integrations (each connects once to the uns) adding a new system requires re wiring existing ones new systems subscribe to existing topics; nothing else changes data is siloed per system all data is available to all authorised consumers integrations are fragile and version sensitive producers and consumers are fully decoupled beyond integration simplification, the uns enables real time visibility any system can subscribe to any topic and receive data as it is produced, without polling digital twin foundation the uns holds the live state of the plant, making it the natural data source for digital twin implementations ai and analytics readiness all data flows through a single point, making it straightforward to connect analytics engines, machine learning models, or agentic ai systems without custom data pipelines for each scalability new production lines, new sensors, and new systems are added by publishing to new topics; no architectural changes are required key terms topic a named channel in the namespace to which producers publish and from which consumers read broker the central message hub that stores and routes published data producer / consumer a system that publishes data to the uns, or one that subscribes to and reads from it isa 95 hierarchy the enterprise / site / area / line / workcell / device structure used for topic naming frequently asked questions how does mantsu implement a uns? with two complementary brokers an mqtt broker for ot machine data and a redpanda (kafka) broker for it business events, both using isa 95 topic names do you buy a uns product for mantsu? no in mantsu the uns is an architectural pattern realized with these brokers and the naming convention, not a separate product what happens when you add a new system to a mantsu plant? it connects once to the namespace and subscribes to existing topics; no existing system has to be re wired why does mantsu use the isa 95 hierarchy for topic names? it mirrors the physical plant structure, making every topic in the namespace self describing and discoverable summary a unified namespace is a single, hierarchical data space through which all plant data flows producers publish once, authorized consumers subscribe, and integration effort grows linearly instead of exponentially enterprise / site / area / line / workcell / device / datapoint