Hard Disk Failure Much Higher than Reported

Rich Tate

Supreme [H]ardness
Joined
Jun 9, 2005
Messages
5,955
Carnegie Mellon University recently conducted a study that found that HD failure rates are actually much higher than reported. If you are in the market for a new drive, I’d suggest giving this one a good read.

The Carnegie Mellon study examined large production systems, including high-performance computing sites and Internet services sites running SCSI, FC and SATA drives. The data sheets for those drives listed MTTF between 1 million to 1.5 million hours, which the study said should mean annual failure rates "of at most 0.88%." However, the study showed typical annual replacement rates of between 2% and 4%, "and up to 13% observed on some systems."
 
Ever since Seagate started the 5 year warranties I won't look at any other brand. If two similar spec drives cost the same money but one has a 3 year warranty and the other has 5 I consider that a no brainer. The fact that Seagate has as good a reputation as anyone making drives pretty much seals the deal. I like a company that is willing to stand behind their product in a way that makes you feel safe buying it.
 
Ever since Seagate started the 5 year warranties I won't look at any other brand. If two similar spec drives cost the same money but one has a 3 year warranty and the other has 5 I consider that a no brainer. The fact that Seagate has as good a reputation as anyone making drives pretty much seals the deal. I like a company that is willing to stand behind their product in a way that makes you feel safe buying it.

The warranty is practically irrelevant. If you have a 500GB drive fail, and it's not under warranty, what's your biggest problem? Is it that you need to buy a new drive or that you may have lost tons of data? I'm going with the latter.

I just had a drive fail, and it's the first time in 18 years of computing that I've had a HD go bad out of warranty. Ironically, the big loss wasn't form that, since I'd backed up some of the data already and moved the rest to a new drive. The big loss was today when the drive I backed up to lost all partitions. The drive is under warranty, but it's not defective (AFAICT) and even if it is, who cares? I lost data, which is worth far more than a $130.00 drive (which will be worth about $20.00 after 3 years).
 
planned obsolescence!!!

and yeah, fuck warranties. too much of a pain in the ass to worry about.
 
I've had one drive replaced by WD about 10 yrs ago. I was a struggling student at the time so it was worth it as I couldn't afford another one then.

And that incident also gave me one big lesson on backing up my data. I had just finished a report that had taken a few days to complete and was going to put it on a floppy to hand in but the drive became toast and so did the report. Stayed up the rest of the night re-writing the report.

Now everything goes onto dvd. If someone loses data because of HDD failure then it is their own fault.

The last line of that story is very true.

"If they told me it was 100,000 hours, I'd still protect it the same way. If they told me it was 5 million hours I'd still protect it the same way. I have to assume every drive could fail."
 
this is one reason why i will never replace my kingwin case till i can find another one with rubber grommet hard drive mounts. it make me sleep better at night, literally because i dont hear my hard drives ticking away and resonating the rest of the case. and im sure hard drive life is related to how much of a beating they take. these grommets seriously do take a lot of the shock off the drive... anyone know of any other nice cases that have this feature?
 
I'm not so sure the article is all that groundbreaking. Think about it for a second....the drive manufacturers are probably testing in an environment where the drives are being used in a "best-case" scenario. A clean environment, dust-free, typical usage with ideal conditions inside the computer case (if they even test them inside an actual computer).

If a real-world environment that drives has passed through at least two other's hands on the way to you. The shipper probably abused the packaging box on the way to you, the tech that installed might not have used proper static protection...drives that fail in the real world have probably gone through hell on their way to dying.

It really doesn't surprise me that real-world usage has a higher failure rate. People don't know how to properly take care of electronics...starting with the wholesalers and ending with the consumer of that drive.
 
I'm not so sure the article is all that groundbreaking. Think about it for a second....the drive manufacturers are probably testing in an environment where the drives are being used in a "best-case" scenario. A clean environment, dust-free, typical usage with ideal conditions inside the computer case (if they even test them inside an actual computer).

If a real-world environment that drives has passed through at least two other's hands on the way to you. The shipper probably abused the packaging box on the way to you, the tech that installed might not have used proper static protection...drives that fail in the real world have probably gone through hell on their way to dying.

It really doesn't surprise me that real-world usage has a higher failure rate. People don't know how to properly take care of electronics...starting with the wholesalers and ending with the consumer of that drive.


Well the article pretty much says that, but the point is that MTBF is supposed to be an estimation for failure in the real-world, not in a clean room. Who cares about it's failure rate in an environment that the drives will never operate in.

That said, my failure rate on drives has been pretty low. Both were maxtor (same model). I think my mother lost a few maxtors about 10 years ago, and I believe those were the same models.
 
I've replaced WD, Maxtor, IBM, and Seagate drives, in and out of warranty. Quality between them is pretty much a wash. Like the google study suggested, these tend to fail in lots. Several systems bought around the same time all having the same drive are failing prematurely, been there, still seeing that.

I don't know about FC myself, we're using SCA in our servers, but they seem more reliable than run-of-the-mill ATA drives to me. Fewer lots going bad anyway, even if individual drive failures are similar. For that reason, I love my 2 Raptors, not a glitch from them yet in three to five years. But two is hardly a study.

All I buy anymore are WD Raptors and Seagate. The warranty may not make up for your lost data, but when you know Seagate needs to pay for a prematurely failed drive for 5 years, wouldn't you think they'd try harder to make them reliable? At the very least, you get a refurbed drive out of the deal where you'd need to buy a new drive otherwise. Some of my customers opt for a new drive anyway to save time, then they can keep the RMA'd drive when it comes in for a spare. If we could chose between paying for better reliability and the "guarantee" of a warranty, I'd choose the reliability, no question. I just doubt we'd ever get that choice, they go hand in hand usually.

Honestly, with drive prices now, why aren't RAID-1 systems in the mainstream yet? Wouldn't that be a good selling point?
 
The warranty is practically irrelevant. If you have a 500GB drive fail, and it's not under warranty, what's your biggest problem? Is it that you need to buy a new drive or that you may have lost tons of data? I'm going with the latter.

If you have a 500GB drive an aren't backing up your important data, you are a FOOL, and that is your biggest problem.
 
I lost data, which is worth far more than a $130.00 drive (which will be worth about $20.00 after 3 years).

So, don't you feel foolish by not having raid, or an external mirror storage? Since it's only $130, after all.


whistles and points to sig. You think this stuff isn't backed up?
 
So, don't you feel foolish by not having raid, or an external mirror storage? Since it's only $130, after all.


whistles and points to sig. You think this stuff isn't backed up?

Say it with me slowly....RAID....is not.....backup! RAID is a downtime prevention technique, not a backup in the cause of catastrophic failure.
 
I've replaced WD, Maxtor, IBM, and Seagate drives, in and out of warranty. Quality between them is pretty much a wash. Like the google study suggested, these tend to fail in lots. Several systems bought around the same time all having the same drive are failing prematurely, been there, still seeing that.

I think this is a reply to me. I can see how it might seem like I was saying Maxtor is less reliable, but I wasn't saying that. Just the opposite. Although all the drives that I have direct knowledge of failing were Maxtors, but in my case they were the same model (the 2nd drive that failed was the in warranty replacement for the first) and in my mothers case, I believe they were Dell or Micron machines and they all had the same HD's. That fits in with some drive models/batches are less reliable than others. They just happened to be maxtors in these cases.

I've heard people say WD's are the worst and a few years back it was Seagate that was crap, but Quantum was king. Personally, I bought my last drive, because it was the quietest large drive (wd5000ks). If Seagate had been the quietest, id' have bought it (after all it did come with a warranty that was 67% longer and they were cheaper too).

All I buy anymore are WD Raptors and Seagate. The warranty may not make up for your lost data, but when you know Seagate needs to pay for a prematurely failed drive for 5 years, wouldn't you think they'd try harder to make them reliable? At the very least, you

Maybe. But maybe they figure the extra insurance costs are made up for by increased sales. What was the typical HD size 4 years ago: 80gb? I had a 160GB, but I think most people bought much smaller drives. For the former, I'd bet the drives are relatively cheap to replace and there are plenty of refurbs they can send out. For the latter, I suspect it was an enthusiast purchase (at that time) and we probably outgrew the drive long before 5 years are up (I know my 200gb and 160gb drives have been on ice in my Athlon XP 1800 machine for 2 years). I use it occasionally when things go to hell (like this past week) or when a friend comes in town and he uses it to play EQ, but not very often. I'm not saying I wouldn't take the free drive -- if it's free, I'll take 3 -- but I'm not sure there's a huge cost to Seagate from home users. For businesses, the extra 2 years is more meaningful, but WD's RE models have 5 years too, so it's not an issue, unless it's a desktop machine, and those probably all came from a Dell or HP.

Anyway, sorry for rambling so long. All things being equal, id' take a longer warranty over a shorter one, but I wouldn't pay much for it and as I said, I chose the quiet drive with 3 years over the less quiet drive with 5 (and arguably a drive with better technology).

Honestly, with drive prices now, why aren't RAID-1 systems in the mainstream yet? Wouldn't that be a good selling point?

I'm thinking about doing that after this past week (though given that I had 1 start to go bad and the other have it's partition tables go up in smoke, I'm not sure it would have mattered). Though I'm not sure if I want to do that or just set up some sort of external storage and back up every night.
 
nilepez, I wasn't actually pointing to you, and, incidentally, I have the least faith in Maxtor's as well. At least all the one's that are pre-Quantum merger. I replaced one on my mother's PC as well, go figure. ;) I even built that one.

About old drives though, especially "enthusiast" drives. I still use them. I've got 5 computers. Only the servers need the space, and the gaming machine needs the speed. I've got 40's, 120's, 250's and 320's now, I always buy $/GB. Usually $100-$150 for the big drives for storage, the Raptors are the exception of course. 40GB drives make good Application/game drives, or scratch drives for encoding, though I've largely moved up to the 120's for that now. The only drive that really needs the speed is the Windows drive, games are streamed. Many performance tests show little improvement in game load times between drives.

Like RAID, I don't understand why everyone wants a single drive. I've got four in this computer, four in the HTPC, three in the server, two in another workstation. I never have just one if I can help it.

Until my last WD250 died, I figured my data would be safe on a drive used exclusively for storage. No churning data, no running apps, no moving, pasting, copying, deleting, and no need to even defragment. Create my movies on a scratch drive, then move them to the data drive. After a year, only the Thumbnail files got fragmented. Windows would routinely idle the drive for lack of use, on a system that ran 24x7 time shifting and encoding video. Then, well, it died. Less than a year of service, still in its' one year warranty even. 200GB lost. Though the movies have since been recreated, that's still hundreds of hours of processing time, not to mention my time spent repeating the process. No real need for backup, I can still pull out the original movie. It's just convenient having them all at my fingertips.

RAID is a good idea if you are largely looking at a single drive anyway and you can set up a simple mirrored pair, or if you have the money to drop on several identical drives at once you could set up a RAID-5. But I have just found it easier, and more cost effective, to migrate data from drive to drive, system to system, and buy one drive at a time. But then, working on my system is playtime for me anyhow. :)

Sometimes I dare myself to crash the system, just so I can have a reason to rebuild it. I don't want to be scared to do it, or mad about it when things go wrong. I want to be able to reverse the setback, and build a better system.

You're not the only one who likes to rant. :D

huh, threads getting old anyway, I'm probly talking to myself anywayz.

Yea self, you talk to much, get outa here!
 
Back
Top