Downtime with a number on it
Manufacturers are unusual among our clients in knowing precisely what an hour of stopped production costs. That makes IT investment decisions refreshingly straightforward — the business case for redundancy writes itself — and it makes getting it wrong immediately visible.
It also means the priorities are different from an office business. Recovery time dominates. A system that is slow is annoying; a system that is down stops the floor.
The unsupported machinery problem
Nearly every manufacturer we work with has at least one machine whose controller runs an operating system that stopped receiving security updates years ago. A CNC machine, a production line HMI, a test rig. The vendor never certified anything newer, and the machine cost six figures and works perfectly.
The bad advice is to replace it. The economics almost never support that, and the constraint is real rather than negligence.
The workable answer is segmentation. The controller goes on its own network segment with tightly controlled traffic in and out, no internet path, and no route to the office network. It remains theoretically vulnerable and practically unreachable. That is a legitimate and defensible position, and it is what we implement.
Separating office from production
This is the single highest-value structural change for most manufacturers, and a large majority of the sites we assess have not made it.
On a flat network, the reception PC, the accounts machine, the guest Wi-Fi and the production line controller share one broadcast domain. A phishing click at the front desk has a clear path to the factory floor.
Segmentation separates them: office, production, guest and connected devices on distinct VLANs with deliberate rules about what may cross. It costs design time on a network you are already running, and it is what limits an incident to one part of the business rather than all of it.
Warehouse Wi-Fi is its own discipline
Office wireless design does not transfer. High ceilings put access points a long way from handheld scanners. Steel racking blocks and reflects. Stock level changes alter the environment through the day. Forklifts move through the coverage.
A survey performed with the racking full is the only reliable basis for placement, and the instinctive fix — adding another access point — frequently degrades performance through co-channel interference rather than improving it.
ERP and the four-hour question
Whatever your ERP, the useful exercise is establishing what the business does while it is being recovered. Can you still pick and ship? Can you tell a customer where their order is? Is there a paper fallback anybody has practised?
Writing that down is cheap and it converts an ERP outage from a stoppage into a slower day. It belongs in the same conversation as disaster recovery planning, and for a manufacturer it is usually the part with the clearest payback.