Unified Namespace: How Metal Companies Escape Their Spaghetti Integrations | Newsletter

Every metalworking company that’s serious about digitalization knows the story. You start with a clear vision, Smart Factory, Industry 4.0, everything connected, and somewhere along the way that dream turns into a tangle of integrations.

In this edition I explain why that happens, and how a Unified Namespace (UNS) helps you structurally get rid of it.

From dream to spaghetti

You start with a clear vision. Smart Factory. Industry 4.0. Everything connected. Dashboards with real-time data. Operators with tablets instead of paper work orders.

And you get to work. New ERP, or an upgrade. An integration with your laser cutting machine. Planning software. A dashboard on top. New press brake, different vendor, another integration.

Somewhere along the way the dream turns into something else.

Your laser connects to your ERP, but also directly to your planning, because that one interface doesn’t work well. From there another line runs to a dashboard. There are integrations via XML, via CSV, via APIs. Multiple vendors, each with their own consultants, who don’t quite understand each other’s systems and point fingers as soon as something breaks.

You have dozens of connections. No overview anymore. Data that doesn’t quite add up. People entering the same information in three places. And despite all those systems, someone’s standing there with Excel on the side to keep things workable.

This is what I call spaghetti.

And the problem is: the more you grow, the worse it gets. Every new machine, whether it’s a milling machine, a plasma cutter, or a welding robot, means more connections. At some point you don’t dare change anything, because you don’t know what will break.

Why this happens

It helps to zoom out for a moment. Look at what a sheet metal shop, machining company, or steel fabricator does every day, the cycle that keeps repeating:

Selling, a customer requests something, you create a quote
Planning, the order comes in, you determine what needs to be made when
Production management, the plan is translated to the shop floor
Producing, machines run, operators work
Warehouse management, material comes in, product goes to inventory
Shipping, the product goes out the door
Invoicing, the bill goes to the customer
Matching, accounting matches invoice to payment

And then it starts over.

Every step in that cycle generates data. And every step needs data from other steps. The question is: how does that data flow through your organization?

At most metal companies the answer is: from system to system. Point-to-point. CRM to ERP, ERP to planning, planning to machine, machine to dashboard. Or worse: via paper, manual retyping, Excel as an intermediate station. (Sound familiar? Then also read 3 essential requirements: paperless, connected, real-time, about what it takes to truly replace those paper work orders.)

That’s how spaghetti emerges. Not through bad intentions, but through organic growth without an underlying structure.

A different approach: the central hub

What if you take a fundamentally different approach?

Instead of systems talking directly to each other, you have them all communicate through one central point. A hub.

Your ERP publishes data to that hub. Your CNC machines publish to the same hub. Your planning reads from it. Your dashboards read from it. No more direct connections between systems. Everything runs through that one point.

This principle is called a Unified Namespace, a central place where the current state of your entire operation comes together. It also means your ERP no longer needs to be the center of your IT landscape, it becomes one of the systems that publishes and reads, just like the rest. (More on that in Why ERP no longer needs to be the center of your factory.)

What a UNS is and isn’t

This is where confusion often arises, so let me be clear.

A UNS is not a product. You don’t call a vendor to “buy a UNS.” It’s an architecture, a way of building. You choose components that fit your situation and set them up in a specific way.

A UNS is also not a traditional database. A database is designed to query data, you ask a question, you get an answer. “Give me all orders from last week.”

A UNS works differently. It’s event-driven. Something happens in your factory, a machine starts, an order comes in, a status changes, and that’s immediately published to the hub. Systems that are interested in that information listen along. They don’t need to ask, they get it automatically whenever something changes.

What you see in a UNS is the current state of your factory, now, this moment. It’s a live picture, not a historical report.

And if you do want to look back?

Then you connect a historian.

A historian is a system that listens on the UNS and writes everything to a time-series database. Every event gets a timestamp. Later you can look back: what was the situation yesterday at 14:32? How did this process develop over the past week?

That combination, real-time status via the UNS, plus historical data via the historian, is exactly what you need for serious analysis. Process mining, pattern recognition, and yes, also AI applications in the factory.

But there’s something else important here: that historian sits outside your other systems. You determine what the tables look like. You determine what the columns are called. You determine what you do with that data. No vendor dictating how your data should be structured.

It’s your data. Local. Under your own control.

How you organize data

A UNS needs structure. You can’t just publish everything randomly.

The standard used for this is ISA-95, and it essentially follows how your company is organized:

Enterprise, your entire organization, the holding or group
Site, a physical location, a factory
Area, a department within that location (think: laser cutting, bending, welding, machining)
Line, a production line or process
Cell, a specific machine or work cell

That gives you a logical path for every data source. For example:

metalfab/utrecht/lasercutting/line-1/trulaser-5030/status

The beauty is: once you’ve defined this for a certain type of machine, you can repeat it. Another laser? Same structure, different place in the hierarchy. It scales without additional complexity.

What this delivers

Three things stand out when you start working this way.

It prevents spaghetti. A new machine, different brand? It publishes to the same hub. You don’t need a new integration project where five parties have to sit around a table. One connection to the hub, and everything that needs that data can read along.

You get real-time visibility. You see what’s happening now, not what was in a report yesterday. That might sound like a detail, but it changes how you make decisions. You can’t improve what you can’t see.

You become independent. Your data, your structure, your choices. You’re not locked into what a vendor designed. If you want to do something different with your data, you can.

Where do you start?

There are different ways to set up a UNS.

You can assemble something yourself with open source tools, the so-called ‘MING’ stack. Free, flexible, and you learn exactly how everything works. For experimenting and first steps, this is a good route.

Once you want to scale up, you move on to platforms specifically built for that:

United Manufacturing Hub, open source, widely used in SME metalworking. This is what I work with myself and what I recommend to start with.

Ignition, enterprise platform with proven scalability, more aimed at larger organizations.

HiByte, DataOps platform for data-driven organizations.

FlowFuse, platform for managing and scaling Node-RED flows in production environments.

A note about databases: there’s a lot of AI hype around cloud databases. Be careful with those unless you’re using them for cloud applications, think of combinations like Supabase with Lovable.

My advice: start small. Pick one machine, one data source. Set up the structure. Build a simple dashboard. Learn how it works. Then the next step.

The UNS as foundation for MES and AI

What I see in practice: companies that first get their UNS in order have a much easier time when they then want to set up an MES system. Your production management connects directly to the hub instead of yet more point-to-point integrations. The same goes for AI, without structured, reliable data, every AI project is mopping with the tap running.