HDD Failure Survey by Google

[ORAC]

See BBC Report (http://news.bbc.co.uk/2/hi/technology/6376021.stm) - So, run 'em hot and run 'em hard.....

[Cypherus]

Appears to hold with experience, I have three Samsungs now approaching ten years old on server work, on 24/7/365 and never cooled em once, only HDD I ever had fail was an Hitachi 200gb which tossed in the towel after twenty seven hours, so I guess Hot & Hard works, but only with the right name on the case.

[vapilot2004]

I have had good luck with both Western Digital and Samsung. Highest failure rates experienced around here seem to be from the Maxtors.

I have a friend and former NASA scientist (she was on the ISS design team among other projects from the sixties through the eighties) who throughout her career was told that she was 'rough' on her fixed disks.

She has replaced two drives on her home computer in the past few years. Someone suggested resonance patterns and the type of data stored. :rolleyes:

I guess it could be true. :\

[PPRuNe Pop]

I have had three failures. ALL Hitachi!

[BOAC]

I agree with the Maxtor vote. NB Although the 'shop' you bought from may only offer a 1 year 'guarantee' on them, Maxtor themselves do 3 (?I think?).

[IO540]

I run four PCs at work 24/7 and two PCs at home 24/7. Most are IDE or SATA.

Also at each location is a www/email server which is a normal 3HGz PC with a SCSI 10,000 RPM HD and that runs 24/7.

Comparing the failure rates on the above, zero over 5-10 years, with failure rates on PCs that get turned off, there is not a shadow of doubt that by far the biggest reliability factor is leaving the thing running all the time.

One can't tell if things fail at power-down or power-up (think about it ;) ) but when they do go, they sure as hell are found to be dead at power-up. Since the earliest days of PCs (c. 1982) I have not had a PC pack up while it was running.

I replace drives when they start to make a noise. Trueimage, and clone the drive onto the new one.

The cost is in the electricity used, probably £200/year at home.

[Wodrick]

One thinks the analogy is fair - when do light bulbs fail ?

[CounterSunk]

I also have had a high failure rate with Maxtors, and thinking back failures do seem to coincide with unscheduled power downs or re-boots. I also have a RAID array of Samsungs (Spinpoints) which have run rock steady, despite a couple of times when the aircon failed and temperatures soared, for the last three years.

[IO540]

Some bulbs fail when they are turned off :) But that's not fair, because you can shake a bulb and see the bits of the filament rattling about the bottom...

I've been doing electronics design for about 30 years. Temperature cycling has always been the #1 enemy of electronics. It's much better these days than say 20-30 years ago, largely due to better chip packaging methods, but it is still the #1 killer.

[Cypherus]

While I am assuming that even Samsungs are not immune from failure, they do appear to be very tolerant of abuse, persoanlly I have never had one fail that I have fitted or used, maybe just lucky but can't see me changing suppliers anytime soon.