PLC + SCADA architecture and redundancy
Layered architecture from field devices to enterprise, where OPC UA fits, redundant servers and networks, and what failover really means for data and alarms.
Layers
A common reference (Purdue model / ISA-95 levels):
| Level | Content |
|---|---|
| 0 | Sensors, actuators |
| 1 | PLCs, RTUs, safety controllers, remote I/O |
| 2 | HMI, SCADA servers and clients, engineering stations |
| 3 | Historians, site operations, MES |
| 3.5 | DMZ between OT and IT |
| 4 | Enterprise IT, ERP |
Traffic crosses levels through defined points (firewalls, data diodes, replicated historians), not through flat networks.
OPC UA in SCADA
- As driver: SCADA is an OPC UA client of PLC servers — one standard interface instead of many vendor drivers, with security and browsing of typed models.
- As aggregation: an OPC UA server (gateway) collects several PLCs and offers one address space.
- As upward interface: SCADA itself is an OPC UA server for MES and analytics.
Check subscription settings and the number of monitored items against the PLC server limits.
Redundancy
| Element | Typical solution |
|---|---|
| SCADA server | Hot-standby pair: primary and secondary run, secondary takes over on failure |
| Historian | Redundant collectors with store-and-forward buffers |
| Network | Ring topologies (MRP, RSTP), two independent networks for servers |
| PLC | Redundant CPUs (S7-1500R/H, ControlLogix redundancy) where downtime is costly |
| Power | UPS for servers and network, redundant 24 V supplies |
Test failover regularly: does the standby have the same alarm state (acknowledgements!), do clients reconnect automatically, are history gaps filled from buffers, are commands blocked during switchover?
Keep it simple
Every redundant component adds configuration that can be wrong. Choose redundancy from a risk assessment of downtime cost, not by default.
Connections
Open in Knowledge Graph →Loading the map…
Sources
Updated 2026-10-02