A network that
never stops the factory.

Yes, and that's the norm: stopping a line to do the work costs more than the work itself. Deployment is done in phases and within maintenance windows, with the old network running in parallel until the new section is laid, tested and failing over properly. The old section is only switched off once the new one has been in service for a while without incidents — never on the day it's installed. This takes more planning than a one-off installation: you have to decide the order of the areas and coordinate the occasional cut-over with the maintenance shift or the night shift. In return, you avoid the real cost, which is almost never materials but lost production hours and orders that don't ship. In a plant that's already running, the question that really matters isn't how long the installation takes but how many hours of downtime it requires — and that figure should be in the proposal, not discovered during the work.

No, and it's a common misconception. What doubles is the backbone cable — the main route connecting the network cabinets — which is only one part of the total budget. Managed hardware and labour don't grow at the same rate, because much of that work is the same whether you install one path or two. What does rise more noticeably is installation time, because two different routes have to be physically run through the building so the redundancy is real, rather than two cables side by side in the same tray. The right comparison isn't a simple network versus a ring network, but the extra cost of the ring versus what a production stoppage caused by a cut link costs. Once that sum is done, the ring almost always pays for itself in a plant where stopping a line has a direct, measurable cost.

Yes, and relative to what it prevents, it's one of the cheapest decisions in the whole project. Without segmentation, the factory network is one big pool: an infected admin computer, a network loop created by accident when someone plugs in a home switch, or simply printer and email traffic all compete for the same path the line equipment uses. Separating it with VLANs — independent logical networks over the same cabling — means an office problem, which is the most frequent and least serious type of incident, has no physical way to reach production. It also lets you prioritise critical traffic when the network is busy. You don't need to duplicate cabling to achieve this: it's configuration on the managed hardware the ring already requires, so the added cost is low compared with what it isolates.

It depends on exactly what stops and how much each minute of downtime costs, so there's no universal figure that fits every factory. The first step is to work out that cost per minute — lost product, idle staff, orders that don't ship on time — because that number, not a reference table, tells you how much redundancy is worth paying for. The second, and more important than chasing a high percentage in the proposal, is that uptime is genuinely measured with continuous monitoring, and that it's clear who responds when something fails and how quickly. A well-designed installation with alert thresholds usually anticipates many incidents before they stop anything, because almost no outage happens all at once. If you need to work out what kind of backup connection suits your plant, backup internet for businesses explains how to decide.

In an industrial building, a camera endures things an office camera never does: airborne dust, temperature swings between the interior and the loading bay, vibration from nearby machinery and, at outdoor entrances, rain and direct sun. That rules out any camera designed for home use: you need a high IP rating — the measure of genuine resistance to dust and water — and, if it's near machinery, a vandal-proof housing that won't shift with the vibration and knock the image out of alignment. The connection also matters more than in an office: in a plant full of motors, frequency drives and metal, Wi-Fi suffers interference and dropouts that a PoE-wired camera doesn't, and PoE carries power and data over the same cable, which simplifies runs over long distances. Wi-Fi vs PoE cameras compares when each option makes sense.

Copper has a practical limit of around a hundred metres per run before the signal degrades, and a large industrial building easily exceeds that between the central cabinet and the furthest production areas or loading bays. Fibre optic doesn't have that problem: it handles runs of several kilometres without losing signal quality, so it's the natural solution for linking distant points within the same plant, not just for connecting separate buildings. What changes compared with an office installation is the physical work: the fibre has to be protected in conduit or trays, kept away from motors and drives that generate electromagnetic interference, and — in a redundant installation — routed along physically separate paths. Designing these runs, and deciding where fibre is needed and where copper will do, is exactly what fibre & networks covers.