from the don't-forget-to-back-your-stuff-up dept.
Consumers [on February 2, 2016] filed a class-action lawsuit against data storage company Seagate, after it had continued to sell a 3TB hard drive model that had an 'exceptionally' high failure rate. The case is based on figures released by data backup company Backblaze, who found that failure rates for the ST3000DM001 were not only far higher than other drives, but also did not display a typical 'bathtub-shaped' failure rate curve. Backblaze's report has since been accused of not representing real-world use. Seagate is likely to adopt this line as it responds to the suit.
Also covered at Tom's Hardware which goes into considerable detail as to how these were consumer drives, used in a 24/7 enterprise environment.
In short, by its own admission, Backblaze employed consumer-class drives in a high-volume enterprise-class environment that far exceeded the warranty conditions of the HDDs. Backblaze installed consumer drives into a number of revisions of its own internally developed chassis, many of which utilized a rubber band to "reduce the vibration" of a vertically mounted HDD.
The first revision of the pods, pictured above, had no fasteners for securing the drive into the chassis. As shown, a heavy HDD is mounted vertically on top of a thin multiplexer PCB. The SATA connectors are bearing the full weight of the drive, and factoring the vibration of a normal HDD into the non-supported equation creates the almost perfect recipe for device failure.
Related Stories
AnandTech interviewed Mark Re, SVP and Chief Technology Officer of Seagate, to talk about plans for upcoming hard disk drive (HDD) technologies.
Although shingled magnetic recording (SMR) lowers write speeds, a number of techniques help reduce the impact, such as banding together SMR tracks into certain zones with perpendicular magnetic recording (PMR) zones covering the rest of the drive rather than shingling, or adding more SLC NAND and DRAM cache. Seagate will be expanding its use of SMR to increase density in client drives, not just "cold storage" drives, but will be using partial SMR/partial PMR and caching in order to mitigate write performance issues.
For the moment, Seagate won't be using helium outside of products for capacity-demanding datacenter customers (such as the Seagate Enterprise Capacity 10 TB HDD). The company can reduce fluid flow forces inside air-filled HDDs using purely mechanical solutions. On the other hand, Western Digital has introduced helium-filled drives aimed at consumers and has a marketing name for its technology (HelioSeal).
[Continues...]
Cloud backup business Backblaze: Failure rates fell for high-capacity hard drives
Just 139 out of 10,000 12TB Seagate drives fail a year, and Western Digital's HGST brand has an even better rate of 51 in 10,000, according to cloud backup service provider Backblaze, which has 104,778 drives spinning in its data centre.
It's not an exhaustive study; the firm listed just four brands in its estate, with models ranging from older ones with 3TB of capacity to newer 12TB drives and some 14TB drives from Toshiba. However, it does provide some data points for the curious.
It has 31,146 Seagate 12TB disks and 1,278 HGST 12TB spinners. The backup firm claimed the best ever drives it purchased were 45 Toshiba 5TB units, none of which had failed. But of course the sample size of 45 is too small for a valid annualised failure rate (AFR).
[...] The next best is a Seagate 10TB drive with a 0.33 per cent AFR from a population of 1,210 drives. That means 33 out of a batch of 10,000 would fail each year.
Previously: Seagate Faces Lawsuit Over Defective Hard Drives
Disk Drive Failure Rates
Backblaze Hard Drive Stats for 2017
Backblaze Publishes Hard Drive Reliability Stats for Q1 of 2018
(Score: 2, Troll) by Runaway1956 on Monday February 08 2016, @01:45PM
I actually used BackBlaze's data, in part, to decide which hard drives to purchase for my own use. Reliability data is important, and you use it when you can, wherever you can find it.
But, I can't see that BackBlaze has a valid case here. "Consumer grade" is consumer grade, no matter what field you're in, no matter what your tools. I don't buy any of my tools at Wal-Mart. I seldom buy tools at the overpriced hardware stores, even if they stock the tools I need. When I need tools for industrial use, I go to known reliable industrial tool suppliers. NorthernTool is one of my first stops, or Chicago Pneumatic, maybe Ingersoll Rand. Buying a consumer grade air compressor at the local Sears store is a perfect recipe for failure. My air compressors run 24/7, about 360 days per year. We have two, if one fails for any reason, the other kicks in. If it fails to kick in, we are looking at tens of thousands of dollars loss, PER HOUR. With enough downtime, when penalties start to kick in, that can turn into hundreds of thousands of dollars loss PER HOUR.
Back in the IT world, the hard drives weren't warranted for this kind of service, there was no documentation to suggest that the hard drives would stand up to such service, and BackBlaze's own data shows that they don't survive.
Suck it up, BackBlaze - you took a necessary risk when available resources dictated that you must take that risk. Now, just pony up for the professional quality tools necessary for your operation. Stop being cheap dicks.
(Score: 2, Informative) by Anonymous Coward on Monday February 08 2016, @01:49PM
BackBlaze isn't making "a case." They have nothing to do with the lawsuit.
(Score: 1, Interesting) by Anonymous Coward on Monday February 08 2016, @01:55PM
What? Runaway let his mouth run away from him? That nevah happens.
I own a bunch of these 3TB drives in external cases and they have been pretty bad. I also have a bunch of the 1.5TB seagates that also have a bad rep and they've had a lot of failures too.
(Score: 0) by Anonymous Coward on Monday February 08 2016, @02:08PM
"The case is based on figures released by data backup company Backblaze"
What, both anons are wrong? Have fun with that vendetta.
(Score: 2, Informative) by Anonymous Coward on Monday February 08 2016, @03:27PM
Both anons are correct. Backblaze is not suing anyone.
The numbers from Backblaze are the proof. It isn't just that the seagate numbers are shit, but that all the drives from other manufacturers used under the exact same conditions are not shit.
(Score: 3, Interesting) by NotSanguine on Monday February 08 2016, @08:43PM
I own a bunch of these 3TB drives in external cases and they have been pretty bad.
I have several of these drives as well (internally mounted) and they all failed and had to be replaced.
Granted, that's anecdotal.
No, no, you're not thinking; you're just being logical. --Niels Bohr
(Score: 3, Interesting) by Hairyfeet on Tuesday February 09 2016, @03:20AM
I won't touch Seagate drives over 1TB at the shop because they fail so often, they are real shit.
BTW if anybody wants to know WHY they are shit here is the skinny, please note that this is from one of the hardware forums and was supposedly put there by somebody that worked on the hardware in question so take it with a grain of salt if you want but frankly it fits the data..
Anyway supposedly when Seagate bought Maxtor they got a REALLY cheap ARM controller, how cheap? Cheap enough you could build 4 of these controllers for the cost of a single Seagate. So whats the catch? The catch is the ARM chip is absolute shite and is VERY sensitive to changes in temp, if it gets even a little bit too warm? It loses its little mind and forgets the exact layout of the drive geometry and I'm sure anybody that knows anything about how HDDs work just said "oh shit" at reading that because what happens when it forgets drive layout? Head misalignment and head crashes.
Remember when Seagate was pushing out all those firmware updates? That was them trying to fix the issue with software but you can't fix shit hardware with software so all those patches really did was delay the inevitable hopefully to just past warranty I'm sure Seagate was thinking, similar to how HP pushed out a patch for those Bumpgate Nvidia laptops that just locked the fans at 100% to try to get the unit to stay alive long enough to get past warranty. During this period there was a list going around of which serial numbers to look for as SOME Seagate drives were getting the older Seagate controller as they hadn't ramped up production of the new shit Maxtor controller, so if you got one of those? All was good. Most of us just said "fuck Seagate" and went with Samsung, Hitachi, or Toshiba, all of which had much better MTBF rates.
So there ya go, why the big Seagate drives are shit. If you want to know why the smaller drives are fine? Remember that the Maxtor chip can't handle heat and those new 500Gb-1Tb single platters drives are VERY compact, it concentrates the heat more so they use the Seagate controller on those as the Maxtor would be overheating so quick you'd be lucky if they lasted a week. As for why they don't just chunk the junk? In a word....money. The Maxtor controllers are soooo damned cheap I'm sure they've crunched the numbers and found that even factoring in warranty replacements its still more profitable to use the shit Maxtors than the good Seagates. BTW this also explains why their "enterprise" drives will probably have better numbers, they use the good controller instead of the Maxtor crapola one.
ACs are never seen so don't bother. Always ready to show SJWs for the racists they are.
(Score: 0) by Anonymous Coward on Tuesday February 09 2016, @10:48AM
I have a few of them, bought at Costco, left in their enclosures, and used lightly. None have yet failed, as far as I know.
(Score: 2) by Runaway1956 on Monday February 08 2016, @01:55PM
True, but, "The case is based on figures released by data backup company Backblaze,"
(Score: 4, Informative) by Nerdfest on Monday February 08 2016, @02:19PM
That's a possible problem with the lawsuit, but nothing really to do with BackBlaze. It's great that they do this, and I really don't see much difference between commercial and consumer use these days. Many people run their computers 24/7, and all the machines I see running Windows have the hard drive thrashing continuously. If I were to make a hard drive purchase, I'd consider BackBlaze's data the perfect resource.
(Score: 2) by Whoever on Monday February 08 2016, @04:27PM
For the most part, the differences between consumer grade and enterprise drives are quite small:
1. Price
2. Warranty (maybe).
3. TLER settings. The TLER settings assume that the drive is used in a RAID array and you would prefer that the drive fails rather than waits a longer time and might return the data.
(Score: 2) by Runaway1956 on Monday February 08 2016, @04:53PM
I disagree with that assessment - try this video: https://www.youtube.com/watch?v=cXPr3l5P5AI [youtube.com]
Yes, the guy in the red shirt has deformed hands, don't let that distract you. (it did me for several seconds)
I don't have the means to measure the effectiveness of vibration dampening, so I've just trusted the "experts" and reviews regarding that. That was one of the things that I looked for in my RAID array.
Near the end, he hits on the fact that rebuilding an array is a dangerous operation. "one hiccup during that rebuild" I believe that in most cases, "enterprise" is vastly superior to consumer grade. There will be exceptions, of course. If you compare the best of the best of consumer grade, to the bottom-of-the-barrel enterprise offerings, you're going to see a lot of overlap.
He also mentions "mission critical" grade hard drives. I certainly couldn't justify the expense for those bad boys, but I did hold out for a good deal on enterprise.
(Score: 0) by Anonymous Coward on Monday February 08 2016, @09:40PM
...effectiveness of vibration dampening,...
pet peeve --
dampening = making wet
damping = removing energy from a system
(Score: 2) by Whoever on Tuesday February 09 2016, @04:31AM
Vendor wants you to buy more expensive products. Film at 11.
(Score: 1, Interesting) by Anonymous Coward on Tuesday February 09 2016, @01:06AM
I remember back when there were no "enterprise grade" ATA hard drives, because all of the server vendors only considered SCSI hard drives to be "enterprise grade". Back then, a 146 GB parallel SCSI drive was many more times expensive than a 250 GB or 500 GB parallel ATA or serial ATA hard drive. But after years of seeing independent "whitebox" server builders use desktop parallel ATA hard drives with RAID controllers, such as the 3ware Escalade 7500 [techreport.com] (which resulted in a giant rat's nest of potentially air-blocking PATA cables in a server case), Western Digital announced a new hard drive series tailored to deliver parallel ATA size at a higher reliability rate: The WD Caviar RAID Edition [wdc.com]. The key features were 1 million hours MTBF, and Time Limited Error Recovery, a firmware modification to prevent a drive from taking too long to attempt to correct a sector error, which might be interpreted by a RAID controller as a disk timeout.
Seagate followed suit, with a variant of the Barracuda 7200.10 hard drive, called Barracuda ES [techreport.com]. This was co-timed with a 750 GB maximum capacity, in the race to terabyte hard drives, which scared anyone responsible for doing remote backups, prior to our current times of plentiful bandwidth. However, users had more reason to be scared, when the Barracuda 7200.11 and Barracuda ES.2 started encountering a high frequency of firmware-bricking [techreport.com]. I remember having to fight with servers that had brand new ES.2 firmware drives that were dropping out of arrays fast enough to trigger multi-drive failures, and to add insult to injury, the firmware update process (which involved booting the server off of a USB key) would sometimes brick a drive being updated, resulting in another array rebuild, another RMA, and more weeks of scheduling the firmware update. Eventually, we started receiving WD RE3 drives as replacements, as even the tier 1 server vendor had had enough with the ES.2 drives.(Astute readers will note that Backblaze's report shows a 25% failure rate for 7200.11 drives.)
As for the suggestion that "consumer grade drives shouldn't be used in servers", we aren't talking about low-power laptop drives that can't take the heat, or drives that put themselves to sleep. Hard disk manufacturers work a bit like car manufacturers, in that they build "platforms", and select components for each of those drive lines. Some of the higher lines of drives that are just a slot or two down from enterprise are most likely using the same spindles and platters, with a different firmware tuning. Maybe they're not using all of the "1 million hour MTBF" components, but they're close. Back in the day, we would see WD2500JD drives that ran a very long service life, even among some of the failures.
Maybe Backblaze is rolling the dice by going a couple of notches short of the "enterprise grade" drives that have a >50% price premium. It's odds that they've accepted, since their software is (hopefully) able to recover resiliently from failures. Or maybe WD and Seagate are charging an "enterprise SATA tax" on what is mostly the same build of hardware. It reminds me of car engines, where sometimes they're intentionally overbuilt, and the consumers who find this out can go build that "500 horsepower 2JZ" engine swap with minimal rebuild work. Hard drive manufacturers don't want that, because their margins are already razor-thin, and they have bigger problems in their market, such as competition from solid state drives for volume purchases from consumers and server builders.
(Score: 4, Insightful) by VLM on Monday February 08 2016, @02:23PM
Typical legal strategy, when all else fails, just blame the customer.
"You're holding the phone wrong"
"Nicads have memory effect"
I'm surprised the lead drinkers in Flint haven't been blamed yet under some twisted form of logic.
Magically the hard drive knows if its being used for a home or business. Because a business doing nothing for 8 hours per day while the boss attends meetings is a much more stressful environment than some gamer PC where the thing cooks to keep fan noise low and gamers never sleep so its continuously seeking 12 hours per night.
It stinks of marketing BS trying to increase profit by selling the same product to two groups, but the group with more money is charged more simply because they're a monopoly and they can do it.
(Score: 0) by Anonymous Coward on Monday February 08 2016, @02:35PM
but the group with more money is charged more simply because they're a monopoly and they can do it
Which I think is the whole foundation of blackblazes monetary strategy when it comes to hard drives. It is a clever gambit and one that has proven out in the numbers. Then on top of that they release the numbers to everyone. It is also clear that segate had some serious issues in the 2-3TB drives. That sort of failure rate should show up in the return numbers for segate at all levels over 10 eyars. Which any competent lawyer will ask for in discovery.
Much like MS on the 360 they were probably taking a bath and hiding it and playing accounting trickery to make it look better. The 360 had a 30-40% fail rate with it approaching 100% in 3 years. Segate is/was looking at similar numbers.
(Score: 2) by Runaway1956 on Monday February 08 2016, @02:42PM
Gamers are still using hard drives? I thought they had migrated to SSD's and/or massive amounts of memory. Those who are still using platter drives are most likely to invest in higher quality drives, with a focus on speed. It would still be a pretty sucky gaming rig, but they are unlikely to opt for cheap consumer hard drives.
(Score: 2) by LoRdTAW on Monday February 08 2016, @04:11PM
SSD'd are still costly vs spinning rust. The usual strategy is to install the OS and games onto the SSD and put everything else on the mechanical.
(Score: 3, Interesting) by VLM on Monday February 08 2016, @04:21PM
I'm not sure it would change the argument. That memory is only specified for home use not 24x7 it'll overheat or get cooties or whatever foolishness, when the reality is the 24x7 ram will be the same parts off the same assembly line just sold for more money to people who have more money.
I know power supplies are sold like this. The Chinese factory makes 1000 watt power supplies, then the American reseller sells them for varying prices based on the 500 / 750 / 1000 watt sticker they put on the device.
SSDs are super cheap like $300/TB now. Doesn't matter, "this SSD is business rated which means it costs 3x as much because we can, thats why, that's the only difference other than we blame the victim for all residential failures".
(Score: 2, Informative) by Anonymous Coward on Monday February 08 2016, @05:00PM
There's another aspect to this. Manufacturing is not a perfect process; not all parts are exactly the same.
So often what manufacturers do to increase yields is something called "binning": they will make a whole bunch of (supposedly identical) parts, then test them all. Using your example, a power supply might pass tests at 500W load but fails some tests at 750W. So it goes in the 500W bin, gets a 500W sticker, and is sold at a lower price.
The "problem" here is that there is likely more demand for the 500W parts precisely due to the lower price. Perhaps only 10% of the parts fail tests at 750W, but 90% of the sales are for the 500W model. So what does the manufacturer do? They simply slap 500W stickers on some proportion of the 750W and 1000W bins, selling them at the 500W price point. This means when you buy the cheap version, you're most likely going to get something which could have been sold as a more expensive version—but not necessarily!
(Score: 3, Interesting) by meisterister on Monday February 08 2016, @06:22PM
Actually, it's more the other way around. For most power supplies that aren't really subjected to any kind of scrutiny (think of all of those millions of office boxes or budget gaming builds), the factory in China builds a 300 watt power supply that's rebadged as 500, 750, or even 1000 watts.
(May or may not have been) Posted from my K6-2, Athlon XP, or Pentium I/II/III.
(Score: 1) by ncc74656 on Tuesday February 09 2016, @11:23PM
Unlike hard drives, there really is a difference between SSDs. The more expensive models write only one bit per cell. Less-expensive drives write two or even three bits per cell to wring more capacity out of the installed memory, but are a bit slower (needing to distinguish between 4 or 8 values written to a cell instead of just 2) and possibly a bit less reliable in the long term.
(Score: 0) by Anonymous Coward on Monday February 08 2016, @04:44PM
> I'm surprised the lead drinkers in Flint haven't been blamed yet under some twisted form of logic.
Do you read Soylent? That's exactly what happened here and it got a lot of up votes!
it's worth noting that Flint unofficially elected the current leadership of the city through decades of bad choices.
https://soylentnews.org/comments.pl?sid=11611&cid=288674 [soylentnews.org]
The People spoke, they got what they wanted... so they should get it good and hard.
https://soylentnews.org/comments.pl?sid=11903&cid=296901 [soylentnews.org]
It is utterly callous shit like that which drives people away from here. It poisons the discussion because at best it sucks all the sane people into defending the obvious. Pretty soon people just get tired of wrestling with pigs and seek out better farms, leaving the original farm to the pigs wallowing in their shit.
(Score: 2) by SanityCheck on Monday February 08 2016, @06:50PM
Curious why are you AC? Do you fear reprisals?
(Score: 0) by Anonymous Coward on Monday February 08 2016, @08:36PM
I am AC to let my posts speak for themselves.
(Score: 2) by LoRdTAW on Monday February 08 2016, @04:57PM
I don't get it either. There used to be a black and white distinction between enterprise/high end and desktop. That was in the days of IDE/ATA vs SCSI. IDE/ATA for the desktop and SCSI for workstations and servers. Now we have all sorts of marketing BS thrown at us for what adds up to the same disk with a different pretty color slapped onto it and a firmware tweak.
It started with the WD colors, black, blue, green. Made sense as you high, mid, and low power classification. Now the marketoids have a slew of pointless classifications. Just try to buy a disk from western digital and they doubled the marketing BS: Blue, Blue SSHD, Black, Red, Red Pro, Re, Re+, Se, Purple, Purple NV, Mauve, Lime Green and Salmon (okay, I made up the last three). The kicker is the drives are now certified for silly things like number of bays for RAID certified devices (that pegged my bullshit meter) or a workload rating for enterprise drives in TB per year with "workloads" as low as 180TB/yr (that blew up my bullshit meter).
My guess is they are trying to weasel out of warranty issues by artificially classifying disks into categories so when you try to RMA a failed disk they can reject your claim because you didn't use the disk as intended. Oh, you used a black drive in a desktop raid array when you OBVIOUSLY should have been using red drives? Too bad. You used two blue drives in a Raid 1 to boot a critical system? Get lost. Sorry sir, we received your WD enterprise disk and found that the drive counted 100TB transferred in just 6 months of use when it was clearly stated that they are only allowed 180TB per year. Hard drive Nazi says: No RMA for you! Meanwhile the only differences are firmware and the colorful fisher price sticker on the case. Screw these bastards.
Thanks to the lack of competition in the hard disk market, It's just WD, Hitachi and Seagate. Hitachi and Seagate are at least sane in their offerings with two or three types of disks. SSD's are thankfully making it easier to build disks. You don't need all this critical manufacturing equipment as you just slap chips onto a board, put it in a case, and sell em. There has got to be at least a dozen SSD makers out there: Adata, Intel, Samsung, Crucial, Hitachi, Mushkin, etc.
(Score: 2) by LoRdTAW on Monday February 08 2016, @06:57PM
Addendum:
Seagate appears to also have the same wonky drive classification.
One other thought that came to mind:
If they are now rating drives to be installed in multi bay enclosures, does installing more than one non "multi bay certified" drive hard disk into my desktop void the warranty? Meaning if I have three WD blue drives in my tower and one dies, does that void the warranty?
(Score: 2) by cmn32480 on Monday February 08 2016, @07:09PM
But as TFA pointed out, Seagate is shit. And in my opinion, they have been shit for a decade or more.
Hitachi is owned by Western Digital [wikipedia.org] as of March 8, 2012.
So we are left with shitty or complicated. For that choice, I'll take complicated and buy on the high end of the scale.
"It's a dog eat dog world, and I'm wearing Milkbone underwear" - Norm Peterson
(Score: 2) by LoRdTAW on Monday February 08 2016, @11:44PM
Forgot Hitachi was bought out by WD. So its now down to to manufactures. Doesnt matter much as SSD's will be taking over soon.
(Score: 2) by curunir_wolf on Monday February 08 2016, @02:48PM
Unfortunately, the WD Red 3 TB hard drives seem to have a similar issue. Even worse, because the Reds are marketed as NAS drives.
I don't have the data that Backblaze has (I wish someone would do something similar with these drives), just review averages from NewEgg. Yea, it's nothing like a scientific sampling, but it tracks fairly well. The failure rates reported on NewEgg for the ST3000DM001 track about the same (worse the normal for hard drives).
I am a crackpot
(Score: 0) by Anonymous Coward on Monday February 08 2016, @05:15PM
That is how seagate mounted the usb3 drives. Resting on usb3 hardware interface to hold the two pieces together. I tossed two of drives 6 months ago after using for 18mos. The 2tb and 4tb horizontal mounted usb3 drives been working great.
(Score: 3, Funny) by Geotti on Monday February 08 2016, @10:33PM
There's a German saying in IT land about Seagate: "Sie geht, oder sie geht nicht" (Seagate: she works, or she doesn't) tagline has been valid for quite a while now.
(Score: 1, Informative) by Anonymous Coward on Tuesday February 09 2016, @12:46AM
I run Spinrite across every drive I purchase, even SSDs. This exercises the drive and allows its ECC to detect and correct most flaws before they become flaws. Occasionally it detects a bad disk at the outset. I can then return it for a full refund or replacement before I've even stored my data on it.
The failure rate for hard drives usually takes the shape of a bathtub curve. The failure rate in the beginning of the drive lives is higher, if they survive that peak, then the failure time falls off and remains low (for several years in the case of spinning media) before rising again towards the end of work cycle... notably: when no longer covered by warranty.
\_/ <- You are here.
(Score: 2) by Hairyfeet on Tuesday February 09 2016, @03:37AM
I'd love to know how you are doing that as Spinrite hasn't had a release since 2004 (hell their front page [grc.com] has a review of their "new" Spinrite 6 from 2004 where they test it on an 80Gb [kickstartnews.com]) and doesn't support drives over 500GB (just throws an error and closes) last I checked.
Sure Spinrite was good back in the day, but today its about as useful as Netscape Navigator.
ACs are never seen so don't bother. Always ready to show SJWs for the racists they are.