We had obviously retired it from one of our desktop PCs when it reached the limits of its capacity and put it into the external case, where it had been working away doing daily back-ups ever since.

According to the manufacturer's specifications, we probably could have gone on using it for decades. The D740X-6L has a mean time to failure (MTTF) rate of 800,000 hours in the field. Let's see now. There are only 8765.8 hours in a year, which means a whole 81 years or so more of faithful reads and writes.

Except we'd be exceptionally stupid to rely on it lasting anywhere near that long. In our opinion, we'd been lucky to get five years out of that drive and it was time to retire it.

Earlier this year, at a conference on file and storage technologies, researchers Bianca Schroeder and Garth Gibson, of Pittsburgh's Carnegie Mellon University, delivered a paper called "Disk Failures in the Real World: What Does an MTTF of One Million Hours Mean to You?".

They concluded that the annual disk replacement rates were much higher than the nominal 1 per cent or less predicted by the MTTF figures used by hard disk manufacturers. Failure rates were generally more like 2 to 4 per cent and, at some sites, as high as 13 per cent.

For drives less than five years old, field replacement rates were larger than what the manufacturers' MTTF datasheets suggested by a factor of two to 10. Between five and eight years old, the field replacement rates are a factor of 30 higher.

Just as disturbing were the findings of a group of Google engineers, Eduardo Pinheiro, Wolf-Dietrich Weber and Luiz Andre Barroso. Google is one of, if not the world's biggest, user of hard drives and the data they collected suggested the range of failure varies from 1.7 per cent for drives in their first year of operation to more than 8.6 per cent in the three-year-old population.

The Google figures also demonstrate that the self-monitoring so-called SMART utilities drive manufacturers release to users on their own are unlikely to be useful in predicting individual drive failures, although drives that throw up a scan error are 39 times more likely than error-free drives to fail within 60 days.

Even operating temperature and activity levels are less reliable indicators, although older drives do have a higher failure rate at higher temperatures.

We suspect a lot of users are relying on older hard drives that are statistically more likely to fail, and that it might be a good idea to check on their age.

And although MTTF figures clearly aren't an accurate measure of longevity, we still regard them as a good comparative measure. So when we decided to install a new drive in the Australian Dragon PC, which we bought recently as our home media centre, we opted for a model with a higher MTTF than the model we recommended for our workhorse PC. It faces a heavy load, constantly writing and wiping very large files.

We chose a 500GB Western Digital WD500YS, which has a MTTF of 1.2 million hours and a five-year warranty. It's designed for server operations, which we decided justified the higher price tag - between $225 and $499, according to internet shopping agents that track IT prices.

The other investment we've made in longer hard drive life is the replacement case for the USB 2.0 external hard drive. The Department of Financial Affairs now has a Welland ME-740PS Greenstar enclosure, which is distributed by Melbourne-based Anyware Computer Accessories.

It shuts itself down when it's idle, saving power and (we hope) extending the life of the hard drive. Slip in an IDE drive (an eSATA version is due shortly), plug in three cables and link it to your PC via the supplied USB 2.0 cable and when it's not being accessed, the drive will auto-switch into idle, suspend or shut-down mode, reducing power consumption to 80, 20 or 7 per cent. It indicates its status with three LED lights. It also supports one-touch back-up.
By reducing the load on our hard drives, we hope in future they will have to find a new departure lounge.