By Charles Swihart, Founder & CEO, Preactive IT Solutions
| Every network outage I have investigated on a production floor had an identifiable architectural cause. None of them were bad luck. All of them were decisions, made or deferred, months or years earlier. |
When a production line goes down because of a network issue, the conversation afterward is almost always about the immediate trigger — a switch that failed, a cable that got damaged, a firmware update that went wrong. What that conversation usually misses is that the trigger was incidental. The real cause was an architecture that had no way to absorb that kind of failure without stopping production.
After 25 years of untangling these environments, I have come to think of network reliability less as a technology question and more as a design philosophy question: was this network built to anticipate failure, or was it built to work until something failed?
HOW PRODUCTION NETWORKS ACTUALLY GET BUILT
Almost no manufacturing network I encounter was designed as a single, coherent architecture. They evolve. A new production line gets added, and a switch gets connected wherever there is an available port. A vendor needs remote access for a new piece of equipment, and a pathway gets opened without much thought about what else it touches. An ERP upgrade requires a new server, and it gets placed on existing infrastructure without redesigning the network around its traffic patterns.
Individually, every one of those decisions was reasonable given the moment. Collectively, they produce a network with no redundancy, no clear segmentation, and no visibility into where the actual single points of failure are — until one of them fails.
THE ARCHITECTURE ELEMENTS THAT ACTUALLY DETERMINE UPTIME
- VLAN segmentation that separates production traffic from business traffic, so a problem on one does not cascade into the other, and so that traffic prioritization can actually be enforced.
- Redundant paths for anything production-critical — meaning a single switch or a single cable run is never the only way data gets from a controller to wherever it needs to go.
- Wireless infrastructure engineered for the facility’s actual RF environment, not a generic access-point layout, since production floors have interference profiles — metal racking, machinery, distance — that office wireless designs do not account for.
- Traffic prioritization (QoS) for production-critical applications, so a large file transfer on the business network cannot starve the bandwidth a real-time control system depends on.
- Proactive monitoring that flags degradation — rising latency, a link approaching saturation, a device intermittently dropping — long before it becomes an outage.
THE COST OF FINDING OUT THE HARD WAY
The financial argument for this kind of architecture is not subtle. Unplanned downtime in a manufacturing environment routinely runs into thousands of dollars per hour once you account for idle labor, missed production schedules, and downstream delivery commitments. A redundant network path or a properly segmented VLAN structure costs a fraction of a single serious outage, but it only pays for itself before the outage, not after.
I have sat across the table from plant managers after an entirely preventable outage, and the frustrating part is never the technology — the fix is usually straightforward. The frustrating part is that the fix was always available, and it simply had not been prioritized until the outage made prioritization non-optional.
WHAT A RELIABILITY AUDIT SHOULD ACTUALLY LOOK AT
A meaningful network reliability assessment for a manufacturing facility should identify every single point of failure on paths that touch production, quantify what an outage on each one would actually cost, and rank the remediation accordingly. That is different from a generic IT audit, which tends to check general health metrics without connecting them to specific production risk. The manufacturers who get this right treat their network the way they treat their production equipment: as a system that requires deliberate maintenance and redundancy planning, not as background infrastructure that only gets attention when it breaks.Â
About the Author
Charles Swihart is the Founder and CEO of Preactive IT Solutions, a process-driven Managed IT Services provider founded in 2003 and specializing in manufacturing, engineering, and construction organizations across Houston, Austin, Beaumont, and San Antonio, Texas. He is the author of On Thin Ice, an Amazon best-selling book on cybersecurity, and was named MSP Titan of the Industry in 2024.
