http://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/ Engadget Comments for Massive Google hard drive survey turns up very interesting things http://www.engadget.com/media/feedlogo.gif http://www.engadget.com en-us Copyright 2008 Weblogs, Inc. The contents of this feed are available for non-commercial use only. Blogsmith http://www.blogsmith.com/http://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/http://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/Feb 18th 2007 10:25PMhttp://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/http://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/Feb 18th 2007 10:25PMhttp://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/http://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/Feb 18th 2007 10:26PMhttp://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/http://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/
sounds reasonable to get these branded HDD for lower temperatures but now after reading this article, im worried about WD HDDs.]]>Feb 18th 2007 10:38PMhttp://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/http://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/
I believe that the articles are referring to cooled hard drives as hard drives that have a dedicated fan system/kept in cooler rooms.

Warm I think would be a good medium between really hot (like my last Maxtor), and really cold (my new Hitachi liquid cooled.) Speaking of which... time to move some cooling ducts around.]]>Feb 18th 2007 11:57PMhttp://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/http://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/>other notable patterns showed that failure rates are indeed definitely correlated to drive manufacturer, model, and age

Okay, someone tip us off. Who knows where Google is spending the most money (brands or models)?

]]>Feb 18th 2007 10:38PMhttp://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/http://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/Feb 18th 2007 10:55PMhttp://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/http://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/
Every HDD producer had a bad day - and only by diversifying one can assure that data will live on regardless of single vendor problem.

As if nobody here had such experience: five Dells shipped to us all had hard drives failed in 2-3 months. (Most ridiculous part: the hard drives were from different vendors, only common part was "Dell"). Or our RAID from Dell failing due to one hard drive failure and then consequent second hard drive failure during populating replacement for first hard drive. (Tape backups are still only way to ensure data survival.)]]>Feb 20th 2007 3:51AMhttp://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/http://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/
Best--
Seagate
Hitachi
IBM
Toshiba
Maxtor
WD
--Worst

I'm going to read the PDF and see if my observations are even remotely accurate.
]]>Feb 18th 2007 11:03PMhttp://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/http://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/Feb 18th 2007 11:52PMhttp://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/http://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/Apr 12th 2007 7:21PMhttp://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/http://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/
Seagate is the best, followed by WD.

Stay away from Maxtors and Hitachi's, they suck.]]>Feb 19th 2007 10:49PMhttp://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/http://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/Reliability, best to worst:
1. Seagate
2. Western Digital
3. Maxtor / Quantum / Hitachi / IBM "deathstar" / Samsung

Most of the dead drives I get to see are plain old desktop IDE drives, and I suppose SCSI drives may be different, since they're usually purchased by a completely different type of customer.

Seagate's purchase of Maxtor frightened me. I hope it doesn't mean Seagate is going to use some kind of Maxtor "build extra cheap drives" technology.
]]>Apr 27th 2007 10:36AMhttp://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/http://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/Feb 18th 2007 11:06PMhttp://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/http://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/Feb 18th 2007 11:46PMhttp://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/http://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/
Like a cooler (all around) hard drive is not better than ones that run a little hot. And that drive failures do result from companies more often than random chance.

This is besides the point, but I swear by Western Digital. I own an array of ten of those babies and none have crashed on my yet. My Maxtors, Samsung, and Seagates, on the other hand...]]>Feb 18th 2007 11:53PMhttp://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/http://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/Personally, I've relied on Seagate drives and they've been fine, though the latest one I picked up does actually sound as though it is grinding my data. Literally. Mortar & Pestle type action.]]>Feb 18th 2007 11:56PMhttp://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/http://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/
My sample size is a lot smaller, but I support 600+ PCs, and I definitely agree on two points: 1) failures correspond heavily to particular models, and not where/how they're used. 2) SMART is a poor predictor of failures**

Though I don't work in a corporate environment, we buy PCs similarly; i.e. we buy the same Tier1 (Compaq/HP in our case) model throughout its life. They'll generally have a specific set of drives they'll spec for that PC model. Failures will tend to come from the same model drives, and within a surprisingly narrow time period.

The worst ones in my memory have been:

WD AC36400/AC26400/64AA, 102AA, 205AA (last 3 are the same family, I think)
Seagate Medalist 6-10Gbish (These might be Conner designs since that's what they look like; they also have SeaShield)
20Gb DiamondMAX (I forget which model)
WD200BB/400BB (the silver and silver/black lids, not the later all-black ones)
DiamondMax Plus 8 (the slimline ones). These are going out left right and center now.

The worst SCSIs have been Seagate Barracuda 18LP and Cheetah X15.

On occasion, we'll catch a drive that's reporting SMART errors before they fail, but they usually let go with no warning. However, SMART is much more effective on SCSI drives. HP/Compaq tracks server drive stats really carefully, and failing drives more often than not WILL show SMART errors.
]]>Feb 19th 2007 12:06AMhttp://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/http://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/Feb 19th 2007 12:22AMhttp://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/http://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/
What is Google more than anything else? No, not a search engine...Advertising Firm! Why would they give free "advertising" by saying company x is the best?]]>Feb 19th 2007 12:42AMhttp://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/http://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/Feb 19th 2007 12:47AMhttp://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/http://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/
Also, who is still offering a 3 year warranty? That could be quite telling.]]>Feb 19th 2007 12:56AMhttp://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/http://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/Feb 19th 2007 1:16AMhttp://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/http://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/
That's IMO, anyway...

Can we get a digest of the Google report Engadget? :D]]>Feb 19th 2007 1:23AMhttp://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/http://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/
>>Maybe it lasts a little longer, but its more than worth it for the much greater quality

I meant to say

Maybe it costs a little more, but its more than worth it for the much greater quality]]>Feb 19th 2007 1:23AMhttp://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/http://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/Feb 19th 2007 2:06AMhttp://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/http://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/Feb 19th 2007 2:22AMhttp://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/http://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/Feb 19th 2007 3:04AMhttp://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/http://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/Feb 19th 2007 3:56AMhttp://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/http://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/Feb 19th 2007 4:24AMhttp://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/http://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/
At the same time, if it turns out - unlikely as it is- that the 2004 model Seagate such-and-such was the best ever - it doesn't matter as these are in production only for a year or two before being replaced by other models.]]>Feb 19th 2007 4:51AMhttp://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/http://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/
At the same time, if it turns out - unlikely as it is- that the 2004 model Seagate such-and-such was the best ever - it doesn't matter as these are in production only for a year or two before being replaced by other models.]]>Feb 19th 2007 4:54AMhttp://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/http://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/
Heres a funny one, ive owned a 8GB Quantum Fireball, and it still working right now in my Linux firewall box. That thing cost 300$ in its days, and from that day ive already had to replace 4 harddrives;

20GB of some brand
80GB Maxtor DiamondMax8
200GB Western Digital (forget what type)
160GB Seagate SATA (they replaced it with a 200GB for free)]]>Feb 19th 2007 8:45AMhttp://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/http://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/
Many of you are wondering why they don't list specific makes and models as unreliable - this paper isn't for you. It's for managers of data centres that already have an installed base of drives, and who want to predict failures. The paper is _not_ trying to create some sort of warm/fuzzy brand recognition in consumers that let's them sleep easier after their latest purchase. Yes, science is objective.

Google could have published the names and models, but the paper states that the drives were up to five years old - where would data centres buy these drives anyway? What sort of predictive power would this give us? Obviously brand name doesn't equate to reliability, and the fact that the results that they have drawn span manufacturers and several model years make them significant, and much much more useful than simply saying, "stay away from brand X", which is how people seem to want to view the marketplace.]]>Feb 19th 2007 9:59AMhttp://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/http://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/http://linuxdevices.com/news/NS2659179152.html

The IA's PetaBox installation comprises about 16 racks housing 600 systems with 2,500 spinning drives, for a total capacity of roughly 1.5 petabytes.

Saikley says Capricorn did extensive testing to qualify hard drives for capacity, reliability, and cost, finally choosing Hitachi. "Although Hitachi does not offer an 'enterprise' or '24x7' SATA drive, our testing found their drives to be as reliable as anything out there, enterprise distinction or not," Saikley said.]]>Feb 19th 2007 11:29AMhttp://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/http://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/
LaCie discs are total junk. (A lot of the them contain maxtor hard drives). I have had two of three die in the last year.

I have since moved all storage to G-Technology"G-Drive/ G-Raid drives they are around 2x expensive, but are stable, reliable and use decent chipsets. I recommend them to anyone that works with graphics/video.

regards]]>Feb 19th 2007 12:02PMhttp://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/http://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/how horrifically bad L**** hard drives are.
The posts are delated soon there after.
Is it not allowed to critcise anything on the
site? Is Engadget just advertorial?

Should we get our pompoms out and only
cheerlead?

A response would be appreciated.

Regards

Rodders
]]>Feb 19th 2007 12:07PMhttp://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/http://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/
http://tinyurl.com/34o8lc

According to this guy, his company is finding that WD makes the most reliable Hard Drives.
I've always used Seagate myself, and never have had a problem.]]>Feb 19th 2007 1:04PMhttp://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/http://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/
I have now Samsung SpinPoint and it works silently. Not as Quantums of old days, but yet magnitude better than Seagates.]]>Feb 20th 2007 4:12AMhttp://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/http://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/http://doxi.ca/a88f]]>Feb 19th 2007 4:35PMhttp://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/http://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/
My problem with any of these studies is that the terms are never defined (“Age”, “Temp” Slightly hot? WTH is that?!) and full histories are not known or impossible to know.

The term age brings to mind questions like
‘How many times was the drive/computer moved?’
‘Was the drive ever disconnected (on the shelf) or was it in constant use?’
‘Was the computer the drive was in a workstation, server, in RAID config, what RAID config, controller-card type, data-only, boot-drive, high-usage environment, allowed to stop when idle, 1U server, power-cycled daily, shutdown-overnight, have a controller-card fail, power-supply fail, reformatted, format type….?’

Otherwise, how do I know where this data fits in my world? (And yes, I know some of these questions were answered in this report.)

FYI
The major cause of HDD failure in our SERVERS (SATA 250GB RAID5 / SCSI 36GBRAID5) thus far has been heat (AC power was turned-off at night in our new location, unbeknownst to us, and we lost 1 drive every couple of days for 2 months until we got the cooling right. We ended up replacing every drive in some of our arrays. The servers that had better airflow in the case design had the fewest failures.

The major cause of HDD failure in our WORKSTATIONS (PATA 10GB-80GB single drive configuration) has been age. Old lower capacity drives (10GB-20GB) that have been used and abused (Moved from computer to computer to and/or office to office) make up most of the failures.]]>Feb 19th 2007 8:20PMhttp://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/http://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/
They notes in the paper how their drives are almost always on and therefore, they have a statistically useless amount of data on power cycles.

Perhaps what makes a difference is temperature changes. Think about it: fast temperature changes can obviously cause big problems (try moving glass from the oven to the freezer). Perhaps the thermal stress from warming up and cooling down every power cycle wears out the drive over time.

This would explain the lack of results from a server farm, but be consistent with common wisdom for drives in PCs, as hotter-running drives would undergo a larger temperature change when warming up and cooling down

Any thermodynamicists or materials scientists know this stuff better than me?]]>Feb 19th 2007 8:31PMhttp://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/http://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/Feb 19th 2007 11:51PMhttp://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/http://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/Feb 20th 2007 7:30PMhttp://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/http://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/Feb 22nd 2007 2:12PM