I think they should lable hard drives with their halflife. On 6/26/07, Geoffrion, Ron P [IT] <Ron.Geoffrion@sprint.com> wrote:
MTBF is a measure of certainty.
Thanks,
Ron Geoffrion 913.488.7664
------------------------------ *From:* kclug-bounces@kclug.org [mailto:kclug-bounces@kclug.org] *On Behalf Of *Billy Crook *Sent:* Tuesday, June 26, 2007 11:45 AM *To:* Jonathan Hutchins *Cc:* kclug@kclug.org *Subject:* Re: Stress Testing Hard Drives
There's nothing quite as reassuring as uncertainty.
On 6/26/07, Jonathan Hutchins <hutchins@tarcanfel.org > wrote:
I think the only thing that stress testing a drive would do would be to move it closer to it's failure point. Either that will be early, in which case it might possibly happen while testing, not quite as early, in which case it will happen just after installation instead of a week after installation, or it will be later in the drive's life - in which case it will just happen a bit sooner than it would have.
About the only use I can see for this would bet to stress test a few examples of a certain model of a drive to failure, and see what the MTBF is.
There are also environmental factors to consider. Testing the drive in an open, bench-configured computer really doesn't give you any information about how it will perform in a closed case sandwiched between two other hot drives. This is one reason that some manufacturer's well intentioned MTBF estimates are inaccurate.
Frankly, throwing it off a high building seems just as informative.
If you can write a pattern to the drive and it passes fsck, and you can repeat this two or three times, that is going to be about as good a test as you can usefully perform. A drive that will function that well is an unpredictable distance from failure. _______________________________________________ Kclug mailing list Kclug@kclug.org http://kclug.org/mailman/listinfo/kclug
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Billy Crook wrote:
I think they should lable hard drives with their halflife.
No hard scientific evidence here, but anecdotal accounts show the new(ish) 500-Gig SATA drives have a *MUCH* shorter half-life than the previous generation of 250-Gig drives. The company I work for (NewTek) is shipping pre-configured video editing systems, and the latest includes a 500-Gig drive. They half-life (ie: blink out of existence) so quickly I had difficulty finding enough units to stress-test the imaging system. Of course, the inventory system indicated there were plenty available... Other items with short half-lives: - - Oscilloscope probes (Why are they called 10:1 probes? Because they ten(d) to one(run) away! <groan>). - - Sticks of DRAM - - Miniature screwdrivers - - Favorite pens/pencils - -- Charles Steinkuehler charles@steinkuehler.net -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.0 (MingW32) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFGgYYwLywbqEHdNFwRAiAxAKDLLboUBoEV8OuHMS/logBX+toJMACePf+r qVexJnwq0wF0dsq2GurxRqg= =mcHG -----END PGP SIGNATURE-----
What happens with the disk drive manufacturers is that they will test the HD as part of the last step in the manufacturing process. They will classify them as to their reliability. The drives with the best reliability results will be sold as drives for SAN products. The drives with reliability in the mid range are used for mid range servers and the drives with the lowest reliability are used for home PC/consumer market. So, if you have a SAN and you go to your local PC store and buy a 500GB SATA drive and put it into the drive cradle for the SAN and try to load that drive into the SAN it will quickly be rejected. If you were to purchase a SAN drive and take the HD out of the SAN drive cradle and use it on your PC/Server/Whatever you will have a much more reliable disk drive. Phil
-----Original Message----- From: kclug-bounces@kclug.org [mailto:kclug-bounces@kclug.org] On Behalf Of Charles Steinkuehler Sent: Tuesday, June 26, 2007 4:34 PM To: Billy Crook Cc: kclug@kclug.org Subject: Re: Stress Testing Hard Drives
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
Billy Crook wrote:
I think they should lable hard drives with their halflife.
No hard scientific evidence here, but anecdotal accounts show the new(ish) 500-Gig SATA drives have a *MUCH* shorter half-life than the previous generation of 250-Gig drives.
The company I work for (NewTek) is shipping pre-configured video editing systems, and the latest includes a 500-Gig drive. They half-life (ie: blink out of existence) so quickly I had difficulty finding enough units to stress-test the imaging system. Of course, the inventory system indicated there were plenty available...
Other items with short half-lives: - - Oscilloscope probes (Why are they called 10:1 probes? Because they ten(d) to one(run) away! <groan>). - - Sticks of DRAM - - Miniature screwdrivers - - Favorite pens/pencils
- -- Charles Steinkuehler charles@steinkuehler.net -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.0 (MingW32) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
iD8DBQFGgYYwLywbqEHdNFwRAiAxAKDLLboUBoEV8OuHMS/logBX+toJMACePf+r qVexJnwq0wF0dsq2GurxRqg= =mcHG -----END PGP SIGNATURE----- _______________________________________________ Kclug mailing list Kclug@kclug.org http://kclug.org/mailman/listinfo/kclug
On Tuesday 26 June 2007 04:56:40 pm Phil Thayer wrote:
What happens with the disk drive manufacturers is that they will test the HD as part of the last step in the manufacturing process. They will classify them as to their reliability.
So 'splain to me, if the only way to tell how long a drive is going to last before it fails is to run it until it fails, how exactly are they testing them? You may hope that the drives certified for SAN have a tighter quality control standard, but it's not reliability testing that's used to classify them. That's done strictly on a sample basis, not on each piece.
Well, they certainly don't run the drives for 1.4 million hours (the MTBF for a Seagate Cheetah NS) to determine if it will fail. What they have is a testing facility where they run read/write tests on large numbers of drives. They then take the failure results of that test and extrapolate it out to get 1.4 million hours or somewhere around 159 years. If you look at the Seagate Barracuda the MTBF drops to 750,000 hours. This equates to only 85.6 years or about half the Cheetah NS drive. Both the same basic technology as far as the disk platter goes. Phil
-----Original Message----- From: kclug-bounces@kclug.org [mailto:kclug-bounces@kclug.org] On Behalf Of Jonathan Hutchins Sent: Tuesday, June 26, 2007 5:16 PM To: kclug@kclug.org Subject: Re: Stress Testing Hard Drives
On Tuesday 26 June 2007 04:56:40 pm Phil Thayer wrote:
What happens with the disk drive manufacturers is that they will test the HD as part of the last step in the manufacturing process. They will classify them as to their reliability.
So 'splain to me, if the only way to tell how long a drive is going to last before it fails is to run it until it fails, how exactly are they testing them?
You may hope that the drives certified for SAN have a tighter quality control standard, but it's not reliability testing that's used to classify them. That's done strictly on a sample basis, not on each piece. _______________________________________________ Kclug mailing list Kclug@kclug.org http://kclug.org/mailman/listinfo/kclug
participants (4)
-
Billy Crook -
Charles Steinkuehler -
Jonathan Hutchins -
Phil Thayer