Hard disk test 'surprises' Google

Three Google engineers (Eduardo Pinheiro, Wolf-Dietrich Weber and Luiz Andre Barroso) put a report together that examined 100,000 commercial hard drives, ranging from 80GB to 400GB in capacity from a wide variety of manufacturers and models, used at Google to store cached web pages and services since 2001. The report concluded that the impact of heavy use and high temperatures on hard disk drive failure may be exaggerated: "Our data indicate a much weaker correlation between utilisation levels and failures than previous work has suggested. We expected to notice a very strong and consistent correlation between high utilisation and higher failure rates. However our results appear to paint a more complex picture. First, only very young and very old age groups appear to show the expected behaviour," the authors noted.

There is a widely held belief that heavily used hard disks are more likely to fail than those used intermittently. It has also been thought that hard drives preferred cool temperatures to hotter environments. The engineers found, however, that hard drives less than three years old and used a lot are less likely to fail than similarly aged hard drives that are used infrequently. The authors of the report speculated that drives which failed early on in their lifetime had been removed from the overall sample leaving only the older, more robust units. The report also noted that there was a clear trend showing "that lower temperatures are associated with higher failure rates. Only at very high temperatures is there a slight reversal of this trend." But hard drives which are at least three years old were more likely to suffer a failure when used in warmer environments.

News source: BBC News

Report a problem with article
Previous Story

Sumatra PDF 0.4

Next Story

Windows Internet Explorer 7 MUI Pack

22 Comments

Commenting is disabled on this article.

I've found in my years of drive ownage that they like being left at a constant temperature. If you power the machine up and down a lot they tend to fail faster. I just leave my machines on all the time and make sure they have plenty of air moving over the drives and keep it at that.

This is just what they found out and shared. Doesn't mean it's true. Although, I'd like to see someone else do a study on 100,000 hard drives.

No it’s just another case of "Google studied the results of hard drive failures on their servers, produced a report and here is the conclusion".

Dakkaroth said,
This is just what they found out and shared. Doesn't mean it's true. Although, I'd like to see someone else do a study on 100,000 hard drives.

Only Google have the time (and money) to do this with 100k drives tbh!

You think they tested 100k drives just for the fun of it? They probably accumulated detailed information about this for years and now just decided to post a report.
I for one would like to know more about the mentioned temps and HDD make and model. I thought big companies always stuck to HP, Dell, IBM, or so. I wonder what commercial HDD they tested and where they're from (where at google).

Doesn't this seem to contradict pretty much every principal in working with hard drives? General rule of thumb being that cooler drive = happier drive = (in theory) less stress/wear-and-tear = (in theory) lower chance of failure. Their report seems a bit fishy smelling, IMHO.

I would guess that there is an optimum tempreature and going below that teperature would have the same ill effects as going above that temperature?

I'm not sure if you know this, but when using the Scientific Method, your general hypothesis is not always correct. Besides, who in their right mind would fudge up a report like this? There's no positive gain for Google in doing that.

It's possible that the lubricants in the motors become more viscoous at lower temperatures, causing more friction and therefore more wear and tear. So much for sticking your HDDs against your IceMaster 5000...

In one of the PocketGamer web servers we used a 40GB Western Digital drive that was thrashed continuously for its entire 5 year life until we replaced the server, and as far as I know is still going!

When we handed the server back I checked the drive's SMART status - it indicated a runtime of about 42000 hours, with just 12 power cycles during that time.

If the test were to be accurate at all, they'd need to have same brand technology. My guess is that all their hard drives are purchased from the same company.

Though, I'd still like to know what company they purchased from. My guess is definitely not Maxtor. :P

Dakkaroth said,
My guess is that all their hard drives are purchased from the same company.

Why guess when the details are clearly stated at the beginning of the article!

QUOTE:
----------
... examined 100,000 commercial hard drives, ranging from 80GB to 400GB in capacity from a wide variety of manufacturers and models
----------

WDGC said,

Why guess when the details are clearly stated at the beginning of the article!

QUOTE:
----------
... examined 100,000 commercial hard drives, ranging from 80GB to 400GB in capacity from a wide variety of manufacturers and models
----------

Ah, missed the word "manufacturers". :O

Though, I would think that if a particular brand were failing more often than others, it would probably be a good assumption that it had to do with the manufacturer.

Co_Co said,
It would be good, if they could tell us what brands failed the least/most and what sizes.

I bet Maxtor is the one that fails the most.
Their HDs are the biggest crap ever, I had nothing but trouble with them :angry:
Maxtor's "support" is abysmal, too

Aero Ultimate said,

I bet Maxtor is the one that fails the most.
Their HDs are the biggest crap ever, I had nothing but trouble with them :angry:
Maxtor's "support" is abysmal, too

Maxtor = bad. That's a very common formula.