7200rpm reliability in 2013 - WD, Toshiba or Seagate?

D

David_CAN

Limp Gawd
Joined
Aug 15, 2011
Messages
186
The last time I was in the hard drive market was just before the great flood, snatching up a pair of great 2TB Hitachis. Those are still running well as a mirrored pair but I'm seeing warning signs with a similar age 2TB Seagate Barracuda XT.

It runs a little hot (high 40s but it's gotten close to 60 under some situations) and I'm occasionally seeing some extra seek time I don't like. RRER is still at worst 1 error in 1 billion seeks but it's been fluxuating up and down a lot. Honestly I've only used Seagate drives a few times in the past and running hot and drive errors seem to be a pattern.

I'm looking at a possible replacement 2TB 7200rpm drive (would be used as a boot volume) and I want reliability, something I can just setup and forget. I can afford a 2TB WD Black drive, though if there is something less expensive that is just as or more reliable and fast for a boot volume I'd be interested.

So is WD the only name that is reliable for non-enterprise drives? Or has Toshiba done something with the Hitachi stuff it got - I really only found one drive from them with mixed comments. Should I avoid Seagate like the plague if I want reliability and am willing to pay a reasonable price instead of taking whatever is cheapest.
 
Google said there wasn't much relationship between HD temperature and reliability. Still, I try to install HDs 1/2" apart and vertically for better convection cooling. Vertical seems to make the aluminum casting 1-2 Celcius cooler, and some power chips run 10-18 Fahrenheit cooler, at least if they face outward (on most drives, they don't).

Going back to the first HD I bought, a 6GB IBM, I've never had one fail, so I get whatever is on sale (but not refurbished or recertified) and test it for several days before putting it to real use. I missed the Seagate 7200.11 fiasco, which was due to bad firmware, but one data recovery company said that it and older Seagates had a common mechanical design flaw that could make its shaft jam up, but the flaw was eliminated starting with the 7200.12.

Often external drives cost less than internal of the same capacity, and last week NewEgg offered a 2TB Seagate Expansion USB 3.0 for $70 that contained a 7200 RPM (as did the one I bought in April). One of my 2TB GoFlex drives had the same, but the other had a 5900 RPM green.

I don't know how good Toshiba HDs are, but their tech support is worse than Seagate's or WD's.
 
Last edited:
Actually, the 2007 google study found that temperature can have a significant effect on HDD failure rates. Particularly low temperatures. Unfortunately, they did not have any drives at much higher than 50C in the study, so there is little data on failure rates for higher temperatures.

lb31XgY.png
 
WD is the best from my experience, seagate are okay, but i feel as if they deteriorate quick
 
First, that graph doesn't make sense to me. 25c is 77f. So are they telling me that drives at 100f are more reliable than 77f? I doubt it

Second, OP, reliability is difficult to judge. Across the board failure rates are on average a few percent for major brand hard drives. When you're only going 1 drive, no matter the brand, it could fail at any point.

Point is, with just one drive, get a good performing drive at a great price. You can't count on reliability no matter what brand.

There is a French retailer that used to post it's return rates on hard drives, which should give a good idea of failure rates.

LINK 1

LINK 2
 
bigdogchris said:
First, that graph doesn't make sense to me. 25c is 77f. So are they telling me that drives at 100f are more reliable than 77f? I doubt it
Click to expand...

Remember, HDs are mechanical. Just like your car engine, it's designed to run at a specific temperature, usually 190F. If you were to yank the thermostat and drive around with the engine at 70F (not that it would run that cool) you'd significantly reduce engine life.

I have no idea what temps modern HDs are designed at, but I'm quite sure it's above ambient.
 
longblock454 said:
Remember, HDs are mechanical. Just like your car engine, it's designed to run at a specific temperature, usually 190F. If you were to yank the thermostat and drive around with the engine at 70F (not that it would run that cool) you'd significantly reduce engine life.

I have no idea what temps modern HDs are designed at, but I'm quite sure it's above ambient.
Click to expand...
Car engines have a very small temperature range for optimal internal combustion. Ambient temperature does not impact internal temperature much on a engine because of the thermostat.

Hard drives on the other hand are supposed to operate at a very wide range of temperatures and ambient will have a significant impact on the internal temperature.

I don't think that analogy works here.

So again the idea that the hotter the drive is the more reliable doesn't seem true to me. I think that a 25c room is much more common than a 50c room, so you'll have a larger sample and more accurate data.
 
Last edited:
bigdogchris said:
So again the idea that the hotter the drive is the more reliable doesn't seem true to me.
Click to expand...

Nobody burns more HDs than Google, hard to argue with their study.

The wider the temp span you design your device, the less reliable it's going to be, and there is still an average or target point where it will be most reliable.

You'd think someone around here would work in the field and shed some light for us!
 
bigdogchris said:
First, that graph doesn't make sense to me. 25c is 77f. So are they telling me that drives at 100f are more reliable than 77f? I doubt it
Click to expand...

I'm not sure why you are so confused. It is a simple graph, and yes, it does indeed show that the HDDs in the study at an average temperature of 25C had a higher AFR than those at 38C.

Anyone commenting on HDD reliability who has not read the 2007 google paper is not well-informed on the subject.

http://www.google.com/url?sa=t&rct=...vPg0CSZPH6HavPX3g&sig2=9p_1Ni-mLBPHwv-EcySklQ

By the way, another interesting concept from the 2007 google study is that for HDDs 2 years old or less, 45-50C temperatures have little effect on the failure rate. But for HDDs 3 or 4 years old, the 45-50C temperatures cause a significant increase in the failure rates (as compared to the 30-35C HDDs). Apparently the younger drives could handle 45-50C temperatures without issue, but some of the older drives had problems with the 45-50C temperatures. The optimal temperature range for google's drives 2 years old or less was 35-40C, but the 3 or 4 year old drives did better at 30-35C.

teH3W0S.png
 
Last edited:
David_CAN said:
The last time I was in the hard drive market was just before the great flood, snatching up a pair of great 2TB Hitachis. Those are still running well as a mirrored pair but I'm seeing warning signs with a similar age 2TB Seagate Barracuda XT.

It runs a little hot (high 40s but it's gotten close to 60 under some situations) and I'm occasionally seeing some extra seek time I don't like. RRER is still at worst 1 error in 1 billion seeks but it's been fluxuating up and down a lot. Honestly I've only used Seagate drives a few times in the past and running hot and drive errors seem to be a pattern.

I'm looking at a possible replacement 2TB 7200rpm drive (would be used as a boot volume) and I want reliability, something I can just setup and forget. I can afford a 2TB WD Black drive, though if there is something less expensive that is just as or more reliable and fast for a boot volume I'd be interested.

So is WD the only name that is reliable for non-enterprise drives? Or has Toshiba done something with the Hitachi stuff it got - I really only found one drive from them with mixed comments. Should I avoid Seagate like the plague if I want reliability and am willing to pay a reasonable price instead of taking whatever is cheapest.
Click to expand...

Come back and edit the OP to include ALL info on the system in question and use. Making recommendations on limited info voids any validity.
 
I just had a catastrophic failure with an array made from Seagate 3 TB drives. These things are really crap. Unbelievable. And that is after a previous array was assassinated by Seagates misleading (and I think deliberately so) communication on the ES.2 firmware problems around startup failure.

I originally planned to never have a homogeneous array ever again, but the options narrowed down. I shy away from WD because of too many firmware problems, I will never use Seagate again. I had gotten a Toshiba earlier to replace the first failed Seagate and I really liked it, not to mention it was faster, too. So I put together a Toshiba array for now. If drives in there fail I'll put in Hitachis.
 
I've never lost an array, at work or at home, in my 15 years doing it. Constant/Automated monitoring, RAID 6 with spares on-site.

Even if I did I still have complete backups.
 
My Seagate 1.5 tb drive from 2010 developed the click of death sound followed by erratic behavior a couple of months ago...needless to say it was a frantic rush to get everything off the drive before loss of data or total failure of drive. I was lucky. Will not buy Seagate again.
 
longblock454 said:
I've never lost an array, at work or at home, in my 15 years doing it. Constant/Automated monitoring, RAID 6 with spares on-site.

Even if I did I still have complete backups.
Click to expand...

Don't do what I did which is turn off the regular scrub. Everything was working fine until a random non-remarkable failure, which then turned bad on me because of uncovered lots of unreadable blocks, each one kicking out that particular drive. Oops.

I did make it without data loss but lots of readonly time.
 
longblock454 said:
This is the most important part, good job having proper backups.
Click to expand...

I didn't invoke the backup yet. I had a triple-fault in a raid6 due to the read errors. A dd copy skipping the read errors and noting their location allowed force assembly of the array and in Linux you can reasonably easy go through a list of missing blocks and find out which file they are in. Nothing important. My important data has true backup and after all syncs are finished I'll do the md5 dance, more out of curiosity. Since I didn't write anything important during crash party time I could also replace the blocks affected by the bad reads by manually recomputing them from the other failed drives but that's a lot of math for raid6 :)
 
You must log in or register to reply here.
Back
Top