Replacing IT

No one likes the idea of getting rid of equipment that works fine, but if you just wait for things to fail before replacing them you risk at least serious inconvenience, and possibly much worse.
Paul Ticher explains how to get the balance right.

A useful life

There are two basic approaches on workstations (see glossary at end) - piecemeal and wholesale. In each case you first have to decide your policy on replacement. How long do you expect a computer not just to keep working, but also to be able to handle up to date applications, current typical file sizes, etc? As developments have slowed in the business field over the past few years this could easily be five years where previously a typical working life-span was three years. Other factors would influence the equation as well: - do you mainly use basic office products or high-power ones involving lots of data or large files? - does your operating system have a higher overhead (Windows XP) or lower (Linux)? - thin client or fat client?

Having decided how long you expect a computer's useful life to be, you then either replace a third, or a quarter, or a fifth, of your computers each year, or you save up your pennies and replace them all in one go (scarier and harder to arrange financially, but it gives you maximum compatibility).

You can take the same approach to peripherals, with the added option of choosing more expensive printers, for example, with a higher duty-cycle, or cheaper ones which will give up the ghost much sooner if they are hammered all day by a large group of people.

Environmental considerations

Whatever you decide to do, don't just chuck equipment in the skip when you replace it. A computer recycling charity may be able to refurbish it and pass it on to someone else, or recover materials before disposing of it safely. (See the Community Recycling Network at www.crn.org.uk)

If environmental considerations are important to you, then the piecemeal approach to replacement makes sense, with workstations (as well as servers) being cascaded down to less and less critical or less demanding roles as they become more likely to fail. You may find that this enables you to keep equipment in use for longer.

Keeping track

The next element is to keep a maintenance log. While your average computer might last five years, some might do six quite happily, while others keel over after four. One way to decide when a computer is ripe for replacement is to check which ones are misbehaving by looking back over all their problems, small or large. In a sense it doesn't matter why one computer often has problems linking to the network, or takes ages to boot up, or whatever. If you haven't got time to locate the fault, just replace it earlier (or possibly swap it out and see if the user has the same problems with their new machine, in which case you have found the source of the trouble).

Don't get caught short

Then you keep a stock of spare parts. Some of these might be scavenged from redundant machines - network cards, for example, keyboards, mice, the odd monitor. Others may be bought for the purpose - a hard disk, perhaps, a CD drive. Also keep a box of bits of network equipment such as switches or hubs. The idea here is that if something does fail unexpectedly (perhaps because you don't have the money to replace as often as you would like to and are trying to keep things going for longer) you can get a machine or a corner of the network up and running again without too long a delay.
In larger organisations you may even want to have a spare laptop which can be brought out in an emergency.

If you don't have space to store spares, or can't afford the outlay, don't forget that many on-line suppliers are cheap and can deliver the following day.

Where the value lies…

This all assumes, of course, that you have fully recognised where the value lies in your IT system. It's not in the hardware - which, as we have seen, is eminently replaceable - but in the data. So the main element in risk management is to make sure that your data is secure. This means investing as much as you can in a robust file server (proper RAID controllers and disk drives, UPS, etc.), AND making sure that everyone stores their data on it, not locally, AND backing up far more often than you think necessary AND keeping recent copies off site. Monitor the system properly, so that if one of your RAID disks fails you know about it and get it swapped before the other one goes as well.

The message is that equipment failure is inevitable, but it needn't cause a large amount of downtime and you can do a lot to minimise the hassle.

Glossary

File server: the computer at the heart of your network, where all your data is (or should) be stored.
RAID (stands for Redundant Array of Inexpensive Disks): a system for storing data on several disks at once, so that if one fails the data is not lost.
UPS (Uninterruptible Power Supply): a piece of hardware which protects your data by allowing time for it to be saved in the event of a power failure.
Workstation: the computer at which a user works.

With thanks to Esther Regenwetter of Blue Lizard for suggesting some additions when this was first drafted.

Paul Ticher is an independent consultant, and can be contacted at paul@paulticher.com January 2006

More Articles