Seagate now offer only one year warranties on some HDD's | Hacker News Reader

Seagate now offer only one year warranties on some HDD's

techpowerup.com

33 pointsosivertsson37 comments

A while back I bought six 1TB Seagate Baracudas.

Every single drive failed within the warranty period; I RMAed them and got replacements. Every single replacement drive failed too. I RMAed a couple of them, and their REPLACEMENTS have now failed. I believe that's a 233% failure rate, and it would probably have been higher if I hadn't given up and just bought new (non-Seagate) drives. And before you ask - all drives were used normally in desktop machines, and they were kept cool[1].

I have no idea how widespread my experience was, but I've heard rumours of very high failure rates in some models of their 1TB drives. If so, I can understand them wanting to protect themselves from a repeat in the future.

[1]: http://www.codinghorror.com/blog/2006/12/hard-drive-temperat...

233% failure rate isn't really a meaningful term. Did some of the individual disks fail more than once? I doubt it.

One industry-standard term you might consider is "mean time between failures" or "mean time to failure".

The MTBF and MTTF has been so perverted by the drive vendors as to be useless. The 233% makes sense as "failures per purchased drive over 3 years". (or whatever period that is)

Seagate is now using "Annualized Failure Rate" on their drives which is MTBF evaluated for the expected number of power on hours in a year (the first year).

The Barracuda Green series gets the new one year warrantee, but boasts a 0.34% AFR in Seagate's specifications. That is on the order of only 1 drive in 100 failing over a 2 year period. Who in their right mind would cut their warrantee to one year if this was really the case?

AFR is fundamentally misleading. The percentages are compounding, and the probability of the drive failing "within this year" is only fixed if you restart your state machine. In reality, people don't look at whether the drive is going to fail "within the next year" but rather "this drive has failed and I've only had it for x years."

MBTF may have been perverted by PR and marketing departments, but as a mathematical concept it is perfectly sound and indeed the ideal measure for failure rates. The solution isn't to make up a misleading statistic and label it as the new "easier to understand metric," while in reality most home users would feel royally screwed if their drive(s) failed within say 3 years of purchase. The solution is to stick to math and statistics and use mathematically-accepted derivations of MBTF that are rigorously upheld and pinned to an honest metric.

I read that heat has yet to be proven to detriment a drive's life or performance within reason. Sure, 5000 degrees would be bad, but there apparently is no evidence that cooling drives matters.

I can't source the article, but seem to recall is was google related with the data coming from their experience with drives in their data centers. It possibly could have come from BackBlaze, though that is a distant second.

I used to run a small ISP, and found that I had a 10% failure rate on average of anything, before it was even used. Order 100 new servers, 10 will be dead on arrival. Order 1000 drives, 100 will be DOA. 100 network cables, all will work, that was about the only deviation.

With drives, the failure rates were better, however, I never allowed drives to be put into use that were not internally "certified". I would use SoftRAID, which is a Macintosh application, though I am sure there are equivalent on Linux and Windows. While this is nothing more than software RAID for the Mac, many of my servers were not Macintosh. I didn't use the RAID capacity of the software, rather using the drive certification feature instead. It takes about 8 hours per 1TB drive to "certify" it. This will run through every sector on the drive, and make sure it is ok. The softwares ability to "predict" that a drive was about to fail, or susceptible to failure, was very good. I would never use a drive that didn't pass.

One handy feature which I did run on non servers with the software, is that it can monitor the drive and tell me that is suspects failure to be imminent. This is not SMART, though SMART was a first line test, the software went deeper and logged aspects of the drive over time.

Once I started doing this, drive failure went to near nothing. I can't really measure failure in percentage or time, as we would outgrow the capacity of the drive, or the server would be taken out of commission for a faster server before a failure ever happened. Those drives were then given to friends or sold. As a result, we never had an in use failure once we started using this method of testing. 42U worth of server space, with a mix of 1U, 2U, 4U, and a few 6U NAS's, plus switching gear.

Do keep in mind, we were a small local ISP, and could take the time to perform these tests. Someone like BackBlaze has their system created in a way to deal with failure as part of the operation, and it would be a waste of time to perform these tests.

Thanks for sharing your story, and an upvote for all the details. But don't you think it's a little bit dishonest to sell perfectly healthy drives that you knew to be on the verge of failure?

The way I read it, they sold the healthy drives that they had outgrown and needed to replace for operational rather than technical reasons.

If you shopped at Fry's you'd have 10% bad network cables, too.

The Google paper you're referring to is here[1]. Their results show that very high temperatures do affect older drives (> 40ºC, 3+ years), but temperature variation within lower and mid ranges shows little correlation with failure rates.

The important result from Google's and others'[2] research is that lifespan for particular drive models is essentially bimodal, with the first peak during the first couple of months - strongly correlated with manufacturing batch, not just make/model - and the second peak after 3+ years due to age related causes. It's possible that people who experience entire arrays of drives failing within the first few months of use have chosen a particular batch of drives that suffer from high 'infant mortality'. One recommendation is to build arrays from drives with different manufacturing batch number or - if the RAID implementation allows it - use completely different makes and models.

[1] http://research.google.com/archive/disk_failures.pdf [2] http://ieeexplore.ieee.org/xpl/freeabs_all.jsp?arnumber=1285...

More than likely, you're destroying the drives yourself. Do you really think this is just all normal?

I would guess that 95% of the people who experience repeated failed drive issues such as yours, are experiencing them because of power issues on your lines (voltage and AC fluctuations), or P/S problems.

The other 5% are getting drives from the same bad lot #.

I strongly suggest that you invest in either a voltage regulator or a power conditioning UPS.

I know, I know! That's the obvious explanation, right? And yet...

1) All three workstations with the Seagate drives have seen constant failure of Seagate drives. They have different brands of motherboard and power supplies, and checking system and HDD temperature shows nothing unusual (ignoring, for the moment, if temperature even matters).

2) All the computers in the office without the Seagate drives - but on the same power circuit - have seen no hard drive failures. At all. And some of the affected workstations had additional non-Seagate drives, including some ancient old IDE drives. None of the "extra" drives failed either.

3) Not long after the first round of failures, we noticed some power quality issues in the office. I purchased a fairly high-end Liebert UPS which solved the power quality issues, but didn't seem to change the Seagate issues.

4) As WD drives were purchased and swapped into the three affected workstations, the failures seemed to stop. Some of the WD drives have now been in service for significantly longer than some of the failed Seagate drives.

Of course, none of that proves anything, but the obvious explanations don't seem applicable. My guess is that my first half dozen drives were from a bad lot #, and the remaining drives were some mix of: * Also from a bad lot # * Damaged in shipping * Were improperly refurbished * Failed due to sheer random coincidence

I'm pretty sure I didn't break the drive which was dead on arrival from Seagate, if nothing else. :-) If you have any other suggestions, I'd love to hear them though. We were using RAID 1, and had backups to boot, so we didn't lose any data, but it was still quite a pain. If there's something I can do to stop it happening again, I'd love to hear it.

I believe (some of) their 1 TB models were known to suffer from what became known as "the click of death" -- the problem was that prominent.

That's technically illegal in the EU, where two-year warranties on electronics are compulsory. This means EU resellers and OEMs will have to be responsible for that second year, or drop Seagate.

That's interesting. Do you have a reference?

The comments on this thread (from yesterday) along with the original article should prove fruitful: http://news.ycombinator.com/item?id=3395577

I used to work for a company that made medical electronics. They pinched money from everywhere they could - staff, customers, shareholders, even their own salespeoples' commissions. Anywhere a dollar could end up in the CEO's pocket, it would. We had a global one-year warranty... except in Europe, where it was two...

Actually the regulation is useless. When you try to execute the warranty, the vendor often argues that you must produce a technical proof that the defect was pre-existing and not caused by improper use. The cost of independent review is often more than the repair. So you should research the seller's behaviour before buying.

Nothing unpredictable here. There are less than three HDD manufacturers left (do you still count Toshiba as HDD vendor?), so what do you expect? No real competition == (no customer care && no discounts && poor quality of goods)

I've been thinking about buying an external hard drive but having some vague idea about the current failure rates, I concluded that, for my data storage needs, the new disks will be practically useless for me without RAID-1. The external disk will store my media and backup my local files already, so I don't want to have a backup of a backup.

This means I'd have to upgrade the disks of my external RAID-1 unit instead, but my old unit only supports 500GB disks officially.

Looks like it's going to be expensive.

For reference, I still had some IDE (EIDE?) harddrives manufactured in the 90's that spun up without a glitch a few years ago when I still had an old desktop box with the appropriate controllers. Ditto for some early-90's SCSI drives that I turned on in the early 2000's, they worked just fine.

I would trade some speed and silence for a lot more robust platters that I could actually trust.

I just had the most pleasant experience RMAing a Western Digital 2TB hd. I purchased it while on a trip overseas a year ago, and I fully expected to be SOL when it began clicking for no reason (it's been a fixture on my desk and most certainly hasn't been dropped).

Lo and behold, a Google search for "WD RMA" took me to their website where they asked for the serial number. It immediately showed it as being purchased "out of region" and "not covered by warranty," except "replacement only."

I wasn't sure I read that right, but I entered my CC details and two days later I had a new drive at my door. Thank you, WD.

To be fair, Seagate's RMA process is similarly painless. The reason I've bought Seagates and WDs is for the warranties and I've had to RMA both brands; both were easy and didn't try to give me the runaround at any point in the process.

If Seagate is only offering 1-year warranties it's simple; I just won't buy Seagate. It really doesn't matter to me.

The trouble I have with the whole warranty idea for hard drives is that the real value in such devices is the data on them, not the mechanics. I'm trying to think of any hard drive that I or any of my businesses have owned in the past decade or more that could reasonably have been returned under warranty other than for being DOA, given the amount of sensitive data that would have been stored on each of them with even a few weeks of use. The only ones I can think of where it would be acceptable (or, indeed, legal) are the company ones that are in a fully encrypted RAID array, which was an experiment we probably won't repeat because of the horrendous performance hit we see on that particular NAS box.

In this sort of context, I wonder how much all these extended warranties were ever really worth. Do some businesses just return failed drives as a matter of procedure, without regard to the possibility of commercially sensitive data leaking because it was still readable with patience and/or the right equipment? Do home users return drives under warranty without realising that all those juicy bank details in their personal accounts spreadsheet might still be accessible to someone who gets hold of the physical drive later, and that such a person might just be buying these things thousands at a time in some foreign country with no goal other than to harvest things like bank details that can be sold?

Seagate wipes the drives they get returned [1]. I imagine other companies have similar processes.

[1]http://www.seagate.com/ww/v/index.jsp?locale=en-US&name=...

I appreciate the effort they go to, and perhaps for typical home users they might consider it an acceptable risk. However, for my business dealings, Seagate's statement there has limited value, because I still have to release the faulty drive into some sort of postal/courier system with any retrievable data intact and trust that it reaches Seagate to be securely wiped. Unless they provide a specialist collection service with the appropriate certifications that they have taken responsibility for destroying the drive securely, nothing that only kicks in when something reaches them via a third party is likely to be an acceptable risk if you're dealing with data protection for credit card data, medical records, etc.

Why aren't you encrypting?

As I mentioned before, in some cases, we do. Obviously if you're dealing with, say, sensitive customer data like card numbers or any sort of legally privileged information, then you have to have robust security in place across the board.

More often, in the particular cases I'm familiar with, we're talking about commercially sensitive information, where we have to weigh up the costs of complete end-to-end encryption by default against the loss of productivity that results when things like big network file transfers slow way down. In a self-funded small company, with limited staff time and limited cash, total encryption is not always the answer.

For example, I think you can reasonably have a policy that encryption of any removable media is required but anything screwed into a rack in the office is biased toward performance, and in fact that is basically what we do in most cases. But it does limit the value of the kinds of warranties we're talking about, because we're not going to just hand over a hard drive that has probably had sensitive company documents on it to some random delivery service.

Similarly, from a personal point of view, I don't consider theft of my home computers to be a very high risk, so I don't routinely encrypt all the hard drives, even though I have some potentially sensitive data like downloaded bank statements and bills on there. Instead, I expect to keep the hard drives for a relatively long time anyway, and then have them securely wiped/destroyed when they are too old/broken to be useful any more.

Just to be clear, I'm not saying these choices are anyone's problem but my own. I accept the cost of the occasional hard drive failure as a natural consequence of these policies. I'm just saying that it means hard drive manufacturers' long-term warrantes are effectively not worth anything to me or the companies I run. I suspect that in practice plenty of other people/small businesses follow similar procedures, and therefore the warranties probably aren't worth much to them either.

If you send it by a secure courier service or even certain types of post I would think that liability no longer rests with you. If damaging information leaks you could then likely sue the carrier and/or Seagate. A likely solution would be to use a vendor that can appropriately wipe the device before sending it to the manufacturer for warranty. There's really nothing else a hardware manufacturer can do on their end. If you're dealing with information that sensitive, some sort of degauss, replace, and recycle is just part of operating cost.

I have copies of all the data I had on there, and my company has strict rules about using TrueCrypt encryption on any external drives.

Timing also seems opportunistic here too. Hard drive scarcity has increased due to the floods and subsequent production issues in Thailand.

In addition to raising prices, like you would in a traditional supply/demand model, this would seem to be a good time to increasing the favorability of warranty terms to the manufacturer given that they're no longer racing to the bottom on price margins with their competitors.

I bought Maxtor back in the day because of their customer service. I also recommended them to everyone I knew. Without at least good customer service I don't know what I'll do. I would pay $200+ for a 250-500GB if I knew the thing would absolutely not fail within 3, maybe 4 years with high usage.

http://web.archive.org/web/19961110081033/http://www.maxtor....

I've vowed to never buy seagate after going though 4 RMA's on 4 hard drives on a raid 10 array that failed one after the other within a month time frame. After talking to a service rep about how the new hard drives were failing within A DAY of use I was told 'rma'ed drives are recertified and cannot be bad' and thus was my fault. If you value your data or your time do not buy seagate, at least in my experience.

the totally nonscientific, based on anecdote, word for a number of years (since they acquired maxtor) is that seagate drives - fast but crap for reliability. which is bad since samsung drives have been reasonably good but just got acquired by them...oh well, back to hitachi.

"Back to Hitachi" is now "back to WD".

I don't know about the warranties, but Seagate is my favorite manufacturer of HDDs and SSDs. I haven't had a single failure yet.

Yep. Even Seagate thinks their products suck. I agree.

Seagate failure rate is ridiculous. Just do a google search for " seagate drive failure rates" for proof. I bought a 256 GB Seagate barracuda back in the day and one morning it just spit read write errors after maybe 3 years of heavy use. I only buy western digital now. ive never seen one fail. If you have a Seagate then keep automated backups or you'll be sorry.

That may be a bit of confirmation bias. Google's massive hard drive study [1] states that there was no significant difference in failure by manufacturer.

[1]http://labs.google.com/papers/disk_failures.pdf

Reply on news.ycombinator.com