Digital Architectures in Manufacturing: from ISA-95 to IT/OT Convergence

Why structure matters more than technology

There’s a conversation that often repeats itself in Italian manufacturing companies when discussing Industry 4.0: people talk about which technology to adopt—a new MES, IoT sensors, a predictive maintenance system—but almost never about what is needed to make them truly work together. The digital architecture is exactly this: not the technology itself, but the structure that determines whether that technology will generate value or become yet another pilot project left unfinished.

Just like the foundations of a building determine what can be built on top, the digital infrastructure of a plant dictates which technologies can be integrated, how effectively, and how well they can evolve over time. Getting this choice wrong doesn’t just mean wasting IT budget. It means ending up, a few years later, with systems that don’t communicate, inaccessible data, and a company that invested in digitalization without reaping the benefits.

What is meant by digital architecture?

In manufacturing, industrial digital architecture refers to the IT infrastructure that defines a digital ecosystem capable of distributing services, orchestrating data flows, and ensuring communication between heterogeneous systems: from field IoT sensors to corporate ERPs. The Scientific Committee of SPS Italia, in its 2024 Position Paper, effectively describes it as the “load-bearing structure” of the entire digital operation: the means that allows the application architecture to express itself and adapt to technological evolution over time.

The practical consequence is simple but often underestimated: a poorly designed infrastructure doesn’t just limit current performance. It reduces the company’s ability to adopt emerging technologies such as industrial IoT, artificial intelligence, and digital twins, with direct impacts on the return on Industry 4.0 investments. As SPS Italia notes, “it is risky as well as short-sighted” not to give proper weight to this structural foundation.

To navigate this landscape, the industry now has several reference models: consolidated frameworks that guide the design and evolution of industrial digital architectures. Each looks at the problem from a different perspective, and this variety is precisely why a compass is needed to choose.

ISA-95: a standard that shows limits

ISA-95, developed in 1995 by the International Society of Automation, standardized for decades the integration between enterprise planning systems and industrial control systems, especially the interfaces between ERP and MES. It worked well in an era when factories were closed systems, machines weren’t connected to the internet, and data was updated in batches. That era is over.

Today, the limitations are structural, not marginal. ISA-95’s rigid hierarchy cannot support the horizontal integration required by modern supply chains: connecting different systems, across plants, across companies collaborating in a value chain. Real-time data management—essential for IoT process control—is another critical issue: the standard was designed for periodic updates, not continuous sensor data streams. And new requirements such as product traceability, energy management, and circular economy fall entirely outside its original scope.

ISA-95 remains a reference point for those managing consolidated legacy systems. But for those designing new plants or aiming to move toward Industry 4.0, it is a starting point to go beyond, not to preserve.

RAMI 4.0 and IIRA: the two most relevant models for manufacturing

Among modern reference models, two have the greatest operational impact for manufacturing companies.

RAMI 4.0 (Reference Architectural Model Industry 4.0), developed by ZVEI and adopted as the official model of the German Industrie 4.0 Platform, proposes a three-dimensional map integrating the product lifecycle, organizational hierarchy, and six functional layers—from physical assets to the business level. The central concept is the Administration Shell: a tool that assigns each machine or physical component a standardized digital identity, making it accessible and manageable within the company’s digital ecosystem. This is what enables a truly interoperable digital twin: every physical asset has its own digital representation, updated and connected to other systems. Security and privacy are not add-ons but integrated requirements from the design phase. For a manufacturing company starting its smart factory journey, RAMI 4.0 is today the most concrete and operational reference available.

IIRA (Industrial Internet Reference Architecture), published by the Industrial Internet Consortium in 2015, approaches industrial digital architectures with a specific focus: how to collect, manage, and analyze data generated by production plants to make more effective decisions. The model systematically defines the characteristics an IIoT system must have to operate in complex industrial environments: operational safety, continuity, real-time data processing, and the ability to integrate with diverse systems. It is the most suitable reference for companies aiming to use production data to anticipate failures, reduce energy consumption, or improve product quality.

The two standards are not mutually exclusive. In complex environments, RAMI 4.0 can guide system architecture and asset integration, while IIRA provides the framework for analytics and decision-making. The responsible organizations are aware of this complementarity: Platform Industrie 4.0 and the Industrial Internet Consortium have formalized a cooperation that includes future system interoperability.

IA agentica governance AI Act processi dati conformità

The real problem for companies: when legacy OT meets modern IT

This is where we reach the point that conferences and white papers rarely address with sufficient clarity. The real challenge for most Italian manufacturing companies is not choosing between RAMI 4.0 and IIRA. It is making a milling machine from 2003, running Modbus or Profibus, communicate with a 2024 cloud system.

This is IT/OT convergence: the integration between corporate IT systems and the control systems of production plants. Two worlds that have lived separately for decades, with completely different logics, priorities, and life cycles. IT thinks in terms of frequent updates, connectivity, and flexibility. OT thinks in terms of operational continuity, reliability, and decade-long life cycles. Bringing these two technological cultures—and the infrastructures that express them—together is the real “open construction site” of industrial digitalization.

The problem manifests itself in several concrete phases. First, data collection: legacy OT machines often cannot natively transmit data to modern IT systems. The most common solution is the introduction of IoT gateways, hardware or software devices that translate proprietary OT protocols (Modbus, OPC-DA, Profibus) into standard IT protocols (OPC-UA, MQTT). The gateway collects field data, normalizes it, and makes it available to higher architectural layers without modifying existing machines.

The second issue is latency. In production processes requiring real-time control—such as inline quality monitoring or alarm management on critical systems—sending data to the cloud and waiting for a response can be too slow. This is where edge computing and fog computing come into play: processing data directly on the device or plant gateway, minimizing cloud interactions for decisions that cannot wait. The cloud remains ideal for complex analyses on large historical datasets and for training predictive models, but not for real-time process control.

The third issue, often the most underestimated, is security. An isolated OT plant has a limited attack surface. An OT plant connected to the corporate IT network and cloud systems becomes a potential cyberattack vector. The NIS2 Directive (2022/2555), implemented in Italy through Legislative Decree 138/2024, has made this issue unavoidable for manufacturing companies with over 50 employees or €10 million in revenue: it introduces concrete obligations for risk analysis, incident notification, operational continuity measures, and—often overlooked—security verification across the entire ICT supply chain. IT/OT convergence is not just a technical challenge; it is now also a matter of regulatory compliance.

RAMI 4.0, with its Administration Shell concept, offers a structured response: giving each physical asset, even legacy ones, a standardized digital identity that makes it accessible and governable within the digital ecosystem without replacing existing machinery.

What to do in practice?

There is no universally correct architecture. The point is to identify the one best suited to your context: the company’s digital maturity, the complexity of the machine park, the priority processes, and the business objectives the digital transformation must serve.

A structured path always begins with mapping the current state: which OT systems are present, which protocols they use, where connectivity gaps exist, and which production processes would benefit most from real-time data availability. Only with a clear vision does it make sense to choose a reference architecture, define where to place computational intelligence, and set IT/OT security priorities.

The most common—and most expensive—mistake is doing the opposite: choosing the technology first and then trying to adapt the infrastructure. Foundations are laid first, not afterward.

Sources:

Agenda Digitale, Oronzo Lucia, “Industrial automation: architectures supporting systems and production processes” (October 3, 2024) https://www.agendadigitale.eu/industry-4-0/automazione-industriale-le-architetture-a-supporto-di-sistemi-e-processi-produttivi/;
EY, Paolo Spazzini and Francesca Lombardi, “Characteristics of IoT architecture in I4.0” https://www.cni.it/images/eventi/2021/II_3_SPAZZINI_LOMBARDI_Caratteristiche_dellarchitettura_IoT_in_I4.0.pdf;
IBM, “What is data architecture?” https://www.ibm.com/it-it/think/topics/data-architecture;
POLIMI School of Management, “Data Architecture, technologies for data usage” https://www.osservatori.net/blog/big-data-analytics/data-architecture-tecnologie-dati/;
SPS Italia Scientific Committee, “Position Paper 2024. Towards a Human-Centric Digital Transformation” (May 20, 2024) https://realtime.spsitalia.it/897/position-paper-2024-verso-una-trasformazione-digitale-umano-centrica.

Mapping the current state, choosing the right architecture, and managing IT/OT security are steps we at Aidia know well, because we work through them every day with the manufacturing companies we support. Our consulting services start with a question: what does your infrastructure look like today, and what is the next step for your company? Tell us about your context.