Key takeaways
Short answer: Point-to-point integration connects each system directly to every other system that needs its data — quick for two or three connections, but the number of links explodes as you add systems, becoming a brittle tangle. A unified namespace publishes all data to one central, structured hub that any system can subscribe to. Adding a new system becomes a single connection to the hub, not a new wire to every other system. The choice determines whether your data architecture scales. See also edge computing vs cloud manufacturing.
Point-to-point connects systems directly: the MES talks to the historian, the historian to the dashboard, the dashboard to ERP, each a bespoke link. For a handful of systems it is simple and direct. But every new system must be wired to every other system it needs, so the number of connections grows roughly with the square of the systems — and each is a custom integration to build and maintain.
A unified namespace (UNS) flips the model. Instead of systems talking to each other, every system publishes its data to one central, structured hub (often over MQTT), and any system that needs data subscribes to the hub. Systems are decoupled — no system needs to know about any other, only about the hub and its structure.
A plant with five systems on point-to-point has up to ten custom connections to maintain, each breaking when either end changes. Adding a sixth system means wiring it to the five that need it — five new integrations. Move to a unified namespace and those five systems each publish to and subscribe from one hub. Adding the sixth system is a single connection to the hub, and it instantly has access to everything already published. The tangle of ten-plus brittle links becomes a clean hub-and-spoke that scales by addition, not multiplication.
The problem is combinatorial. With N systems, point-to-point can require up to N times (N minus 1) over 2 connections, each a custom integration that breaks when either end changes. At a handful of systems it is manageable; at a dozen it is an unmaintainable web where one change ripples everywhere. A UNS replaces that with N connections — one per system to the hub.
1. Point-to-point that quietly grows into a tangle. Each new link seems small until the web is unmaintainable.
2. A UNS with no data structure. A hub of unstructured tags is just a different mess.
3. No system of record. Publishing without defining the authoritative source of each value.
4. Migrating everything at once. A UNS is best adopted incrementally, system by system.
OEE depends on machine data reaching the OEE system reliably. Under point-to-point, every new line is another custom integration; under a unified namespace, the OEE system simply subscribes to the hub and new lines appear as they publish. The architecture decides how easily OEE scales across a growing plant.
See how Fabrico captures this automatically on your lines — explore OEE for manufacturing or book a demo.
A central, structured hub where all systems publish data and any system can subscribe — decoupling systems from each other.
The number of custom connections grows roughly with the square of the systems, becoming an unmaintainable tangle.
Once you are regularly adding systems or scaling across lines and sites.
Yes — an unstructured hub is just a different mess; the data must be self-describing.
New lines publish to the hub and the OEE system subscribes, instead of building a custom integration each time.
Login
Blog
