Everyone’s talking about Industry 4.0. The integration of the OT and IT worlds sounds great on slides. The board gets to see real-time charts. Analysts are rubbing their hands with glee at the prospect of terabytes of machine data. In practice, however, it’s often a brutal clash between two completely different worlds.
Integrating production with business is a necessity today. There’s no getting around it. But cutting corners is asking for trouble. We need to extract data from closed control systems. We need to feed this data into ERP systems and analytical tools. The key, however, lies in how we go about it. The integration must be strictly controlled.
Clash of civilisations. Stability versus constant renewal
Let’s agree on one fundamental premise. Automation engineers and IT engineers have completely different objectives. For traditional IT, flexibility is key. Operating systems are constantly changing. Security patches are released every week. For an OT engineer, however, only one thing matters: continuity. The machine must keep running. Ideally, for a decade without a single restart.
When these two worlds collide without a proper buffer, sparks fly. Let’s imagine a simple scenario. The IT department releases a night-time update for the antivirus software. Suddenly, the SCADA server on the production floor slows down drastically or needs to be restarted. Communication with the machines breaks down for a moment. The production line grinds to a halt, and the company loses a lot of money.
That is why integrating OT and IT is not simply a matter of running a network cable. Above all, it is a matter of striking the right balance. It involves recognising that business systems cannot, under any circumstances, dictate the conditions on the production floor.
How do you do it properly? Separation and the golden rule of the single track
Proper integration is like a well-designed airlock. Data from the production floor must reach the office. However, it should never flow there directly. We first route it through a secure buffer. This could be a SCADA system, an industrial historian or a dedicated IoT gateway.
This is where the data is organised and filtered. Only in this clean, processed form is it sent upwards to the ERP systems. This ensures a strict separation of roles. Control and critical response times remain entirely at the bottom. The IT systems utilise the information, but have no physical means of interfering with the process from the outside.
This is where the DMZ comes into play. The demilitarised industrial zone is the very foundation of a modern facility. It separates the deterministic world of machines from the dynamic office network. This is simply something that cannot be overlooked in any serious project.
Cybersecurity. When an email from the office brings the factory to a standstill
The absence of a DMZ and direct network connections are an open invitation to hackers. Imagine a very realistic scenario. An office worker clicks on a fake invoice. Ransomware instantly encrypts their hard drive. If the office and production networks are one big, flat network, the virus spreads further.
Within minutes, it infects the HMI control panels. It cuts off communication with the PLCs. The entire factory comes to a standstill. This isn’t a scenario from a Hollywood film. It is the harsh and costly reality faced by many poorly integrated plants in recent years.
Security must be built into the architecture from the very start. It cannot be effectively tacked on at the very end of an integration project. Every interface between OT and IT must be designed with the worst-case failure scenario in mind.
Cloud storage: the sin of data overload
Another major mistake is sending absolutely everything to the cloud. Managers, caught up in the hype surrounding Industry 4.0, often lose all sense of proportion. They feed every voltage spike and every minute vibration of a motor shaft into central analytics systems. They do this without any plan.
The result is always the same. Business analysts are drowning in a sea of information. The network infrastructure is choking on millions of useless data packets. Storing this data in the cloud costs more and more each month. There is no real business value to be found in any of it.
This is simply a way of generating costs under the guise of innovation. Instead of churning out terabytes of junk data, we need to filter that data. The overarching system needs to know that the machine has produced a hundred parts. It doesn’t need to know about millisecond fluctuations in the current on the inverter during each of those cycles.
Edge computing sorts out this mess
This is where edge computing comes to the rescue. Instead of sending raw signals to a server in another country, we analyse them right next to the machine. The industrial computer processes this data on-site, right on the production floor.
It draws specific conclusions and only then sends them to IT. The message is short and to the point. For example: “Performance has dropped by 5%” or “The motor is vibrating; order a bearing for next week”. This drastically reduces the load on the network and greatly simplifies the integration of the two systems.
Most importantly, however, this ensures that the machine retains full operational autonomy. Even if an excavator outside the site cuts the fibre-optic cable and the plant loses its connection to the cloud, the local logic continues to function. The production line carries on, and the company continues to make a profit.
Summary
The integration of OT and IT is not just another IT project that can be handed over to interns. It is a strategic operation on the entire factory’s open nervous system. It only makes sense if we build it on a strict separation of layers, intelligent buffers and absolute control over what we send upstream. A well-executed integration provides the company with powerful insights and a competitive edge in the market. Done amateurishly, on a shoestring budget and by taking shortcuts – it leaves the plant vulnerable to serious network failures and painful cyber-attacks.
