- Subject
- Windows & PowerShell
- Filed
- Source
- [email protected] (Michael Becker)
- Loans
- 837
IT/OT Monitoring: Why Plant Networks and Servers Need Visibility
Unmonitored switches and servers can obscure the cause of production stoppages. Start OT infrastructure checks with switch temperature and server health.


When production stops, a SCADA alarm may show the process failure without revealing what happened underneath it. A switch, server or failed application update can be the trigger, yet those systems may sit between the teams that monitor plant equipment and the teams responsible for business IT. That gap can turn a fault into hours of searching.
That was the focus of a live episode on the 4.0 Solutions channel on November 5, 2026. Host Walker Reynolds spoke with Eric Eckman, OT Channel Manager, US at Paessler, and David Montoya, Head of Operational Technology at Paessler, about why operational technology infrastructure is often left out of monitoring.

The gap between process alarms and infrastructure faults
Controls engineers commonly monitor PLCs and HMIs, while IT teams watch business networks. Industrial switches and firewalls on the plant floor can fall between those responsibilities. If nobody has clear ownership of their condition, faults may go undetected until production is affected.
Reynolds estimated that OT infrastructure monitoring takes place in only 1 to 5% of cases. He contrasted that with IT, where infrastructure monitoring is standard practice. Eckman said he had repeatedly heard that downtime came from switches or servers more often than expected while selling fault-tolerant computers for control rooms.
The practical problem is visibility across layers. A process system can report that a line stopped, but that alarm alone may not tell responders whether a PLC failed, a network connection dropped, or an upstream server stopped supplying the information the plant needs. When teams see only the equipment within their own remit, the first response can focus on the wrong part of the system.
A failed update affected nine plants
Montoya described a proof of concept at a large car manufacturer where nine plants went down. Engineers spent almost 12 hours checking PLCs and the automation stack before locating the cause: an application update had failed on a level 3 server in the Purdue model. That server sent production orders to the plants.
The server had been overlooked because it sat in the space between IT and OT. Montoya said a SCADA alarm identifying the server would have directed the investigation sooner. Reynolds put the average cost of downtime at $120,000 per hour; on that estimate, nearly 12 hours represented a substantial production cost. The account illustrates why monitoring the infrastructure that supports process systems can matter as much as watching the process equipment itself.
Bring infrastructure readings alongside process data
Infrastructure readings become more useful when teams can relate them to process conditions. PRTG has sensors for OPC UA, MQTT and Modbus. Its PRTG OPC UA Server extension can send monitoring data from the OT network into supervisory and control systems.
That arrangement also supports work on a Unified Namespace, where machine and infrastructure information can be brought together. If a team wants machine learning to find relationships between switch health and poor production outcomes, those data need to be available in the same place. The episode’s discussion presents shared access to both kinds of readings as a prerequisite for making those comparisons.
Begin with switches and servers
Montoya’s suggested first check is the temperature of switches connecting production cells. Dust can restrict airflow through fans, causing switch temperatures to rise. A malfunctioning switch can then disrupt network access for PLCs. Tracking temperature gives a team a straightforward signal to investigate before a connection problem becomes a process interruption.
For teams wary of running a broad discovery across OT, the suggested starting scope is limited to the IP addresses of servers and switches. Early scans may reveal issues such as low disk space or TLS certificates that expired 300 days ago. These findings can expose maintenance needs on systems that otherwise receive little attention.
Reynolds also described testing a development board and finding a Raspberry Pi driver polling his PLC so frequently that messages were dropped. It’s a reminder that monitoring and connected software can affect the systems they observe. Teams should understand what devices are communicating with control equipment and how often, especially when investigating dropped messages.
Air-gapped environments add placement and installation questions. The episode points to guidance on deploying PRTG without internet access, including where to place the core server and probes. The full recording runs for about 80 minutes and also covers air-gapped networks and whether teams should build their own monitoring with AI.
For an initial review, check whether plant-floor switches and the servers that feed production systems have clear ownership and visible health signals. Those checks can help responders distinguish a process fault from an infrastructure problem sooner, and give IT and OT teams a shared starting point when production stops.