From OPC DA to OPC UA
The original OPC (OLE for Process Control) specifications, especially OPC DA, solved a real problem in the 1990s: every device had its own driver, and SCADA systems had to talk to hundreds of them. OPC DA standardized data access over Microsoft COM/DCOM — which worked, but tied the standard to Windows and made configuration painful across security boundaries.
OPC UA (Unified Architecture) is a complete rewrite: platform-independent, secure by design, and capable of describing not just values but structure. It is the standard that industrial data exchange has been moving toward since its release in 2008.
What OPC UA actually provides
- Transport independence: runs over TCP (port 4840), HTTPS, and even MQTT for pub/sub scenarios.
- Security: application authentication, user authentication, and message signing/encryption are built into the specification — not bolted on.
- Rich data: values carry timestamps, quality, and engineering units; servers can expose methods and events, not just tags.
- Information modelling: the address space is a typed, object-oriented graph rather than a flat tag list.
- No dependency on Windows: servers and clients exist for embedded devices, Linux, and cloud platforms.
The information model
The key conceptual shift from DA to UA is the address space. A UA server exposes nodes — objects, variables, methods, and types — connected in a hierarchy. An engineer can browse the server and discover not only Tank1.Level but also what a "Tank" is, what units its level is in, and which methods it supports.
This matters because a UA client can adapt to new devices without custom configuration. The same server can expose classic tag data for a historian and a fully modelled asset structure for an analytics application.
Client/server and pub/sub
OPC UA supports two communication patterns:
- Client/server: a client connects to a server, browses the address space, and subscribes to data changes. This is the everyday pattern for HMI/SCADA and historian connections.
- Pub/sub (UAFX/OPC UA FX): publishers broadcast messages to subscribers, typically over MQTT, enabling large-scale, one-to-many distribution without per-client sessions. This is the pattern for IoT and cloud data pipelines.
The two patterns coexist: use client/server for control-room integration and pub/sub for scalable data collection.
Adopting OPC UA
A pragmatic adoption path: expose PLCs and instruments through a UA server (many modern devices have one built in, or use a gateway), integrate HMI/SCADA and the historian over UA, and use pub/sub for edge-to-cloud streaming. Standardizing on UA removes dozens of proprietary drivers and makes your data model — not your protocol stack — the thing you maintain.