Contact Info
Unit 9, 17 Power Ave
Alexandria, NSW 2015
info@hexflow.com.au
+61 (2) 8529 4085
Follow Us
IT support technician at a laptop with diagnostic panels on screen

IT Support Wetherill Park: Mixed-Age Hardware, One Cupboard

The estate was not built once. It was built, extended and re-fitted in three different decades, and most units are still running machines, cabling and network gear from all three at the same time.

IT support Wetherill Park callouts rarely begin with a fault nobody can explain. They begin with a nine-year-old desktop, a server sitting in a cupboard beside the roller door, and a business that cannot stop taking orders while somebody works out which of those two is the problem.

Three Fit-Outs, One Address

The oldest units run grey Cat5 in surface trunking with no labels on the patch panel. The middle era is Cat5e into a wall rack somebody installed properly and nobody has touched since. The newest is a fresh lead-in and a mesh Wi-Fi kit bought from a retailer because coverage was poor.

None of that is a fault until the three are expected to behave as one network — an older switch feeding a newer line, and no record of which port feeds which desk. That is why a small fault takes half a day instead of twenty minutes.

The Cupboard Beside the Roller Door

In a unit with no office the server has to live somewhere, and the cupboard beside the roller door is the only place with power and a lock. It is also the hottest, dustiest corner on the site, it shares a circuit with the door motor, and it usually has no UPS.

Three things are worth checking before anyone quotes a repair:

  • Temperature. Leave a cheap logger in there for a week. A cupboard peaking above 35°C in summer is shortening the life of every drive and power supply in it.
  • Power. If the door and the server share a circuit, every door cycle is a dip. A separate circuit and a line-interactive UPS removes a whole category of mystery reboots.
  • Dust and airflow. A sealed cabinet with a filter beats an open shelf. An open shelf with a fan pulling yard dust through the case is worse than nothing.

The tell

If a machine reboots when the roller door opens, the fault is not the machine. No amount of drive testing will find it, because there is nothing wrong with the drive.

What Actually Fails First

On mixed-age industrial hardware the failure order is fairly predictable, and it is rarely the thing the operator blames:

  • Hard drives past five years. The commonest single cause of a machine that will not boot on a Monday. An SSD clone while the drive is still working costs a fraction of data recovery.
  • Power supplies in a hot cupboard. Cheap units fail quietly and take the mainboard with them.
  • Network switches with no surge protection. A single spike through the incoming line can take out the switch and the port on every device attached to it, which is why several machines “die” on the same afternoon.
  • Wi-Fi in a long shed. Racking full of stock absorbs signal. One access point in the office does not cover the back of a 40-metre shed, no matter what the box said.
  • Anything on USB. Label printers, scanners and scales are usually the first devices to drop when a USB hub or a cable is marginal.

Repair or Replace, on a Number

The decision should be arithmetic, not sentiment. The usual split looks like this:

Machine Age Decision
Office desktop or laptop Under 4 years Repair — usually a drive, memory or power supply
Office desktop or laptop Over 7 years Replace unless it is already on an SSD and only runs email
PC bolted to a machine tool Any Repair while the control software still runs, because replacement means re-commissioning the machine
Server in the cupboard Over 6 years Plan the move to a small cloud or hosted system before it fails, not after

The one exception worth paying for: a machine running a proprietary control or label system. An old PC is cheaper to keep alive than to replace, because the real cost is the software licence and the day somebody has to re-teach it.

Where Support Here Does Not Pay

Two people with laptops and no server do not need a maintenance contract. Remote support as needed plus a backed-up cloud folder covers it, and on-site attendance should stay the exception.

It also does not pay to keep a machine alive because it is familiar. A seven-year-old desktop that costs $400 to repair and fails again in four months has cost more than the $700 replacement before the staff downtime is counted.

What we build

We start with an inventory of what is in the unit — machines, ages, the cupboard, the cabling, what the network looks like now — because every decision above depends on it. Then we fix the cupboard and the power, replace what is failing, and leave a labelled patch panel behind us. Remote first where it can be done remotely, on-site when a cable or a dead machine needs a person.

If you want a second opinion before replacing anything, that is usually a short conversation. Call (02) 8529 4085 or send us the details.

Articles on AI automation for Australian practices by Shahadat Hossain, automation specialist at Hex Flow in Sydney.

Post a Comment