back to article If you work on Seagate's performance drives, time to find another job

As expected, Seagate's revenues totaled $2.6bn in the three months to April 1, the third quarter [PDF] in its fiscal 2016, down 21 per cent year-on-year. And as anticipated, it made a $21m loss – its first loss in many many quarters and way down from the year-ago's $291m profit. On Friday, as the results for Q3 were published …

  1. Steven Jones

    A death and decline so easily forseen...

    The death of performance HDDs is surely one of the most predictable trends of recent times. It was always going to start gradually, but as the price per GB of SSDs inexorably dropped, that gradual trend was surely going to turn into a catastrophic decline. HDDs were simply never going to get remotely near SSD latency figures, and reduced latency is possibly the key performance issue in virtually every area of IT, whether it's processors, memory, storage or communications. HDDs had pretty well hit the physical limit on that issue several years ago.

    It was a bit like Kodak and its relationship with digital photography. Rather than being bold enough to make a full-on transformation, it got stuck half way afraid it might cannibalise it's film income. It's very possible that Kodak wasn't going to survive such a radical change in technology (they did, after all, pioneer some digital sensor technology), but the lesson seems to be if you are in denial over what is an inevitable technical trend and are bold enough to make the required radical decisions when the going is good, you aren't going to be around in the long term.

    That's disruptive technologies for you. Every major tech company needs a strategy for what they might do should their particular golden goose's eggs become irrelevant.

    1. Joerg

      Re: A death and decline so easily forseen...

      There is no death of Hard Disk Drives and SSD.

      It is all b*ll and lies. Smoke and mirrors. Nothing else.

      1. Frumious Bandersnatch

        Re: A death and decline so easily forseen...

        There is no death of Hard Disk Drives and SSD

        Sure, hard drives aren't going away for a while, but there's this thing called "opportunity cost". Seagate seems to have chosen to stick with spinning disks over SSD. In so doing, it's devoting its limited resources to chasing a shrinking market at the expense of building expertise, capacity and market share in the newer SSD market.

        I can only guess that Seagate execs imagine SSDs to be not quite there yet and consider a shift in focus to them being a more risky proposition than riding out the cash cow for a while longer. Maybe they're right, maybe not. Time (and timing) will tell.

      2. asdf

        Re: A death and decline so easily forseen...

        >It is all b*ll and lies. Smoke and mirrors. Nothing else.

        The only people that hold this view long term (sorry if missed sarcasm) have never seen Windows 10 (yuck I know but still) boot up in 5 seconds on a decent quality SSD and or never owned one. We have probably already hit peak spinning rust imho.

    2. a_yank_lurker

      Re: A death and decline so easily forseen...

      Following up on Kodak and film, the first digital cameras could not provide images as good as film. But in many instances even then they were superior. Digital technology was posed to rapidly improve to now digital images are often better than film.

      As a parallel, several camera manufacturers jumped into the digital camera market and successfully transitioned. Nikon and Canon are still going strong especially with their DSLRs.

      1. Joerg

        Re: A death and decline so easily forseen...

        You are seriously trying to compare analog cameras on film to Hard Disk Drives ? Really?

        You have no clue what you are talking about.

        And then SSD NAND would be the new digital camera?

        And so what would the 3D XPoint SSD be?

        Please...

    3. Solmyr ibn Wali Barad

      Re: A death and decline so easily forseen...

      Difficult times indeed. And their quality issues in last 7-8 years haven't been exactly helping. Enterprise drives, nearline drives, consumer drives - all suffer from premature media wear and nasty firmware bugs. About the only saving grace was that WDC/Hitachi chaebol had troubles of its own and could not produce enough kit to wipe the floor clean.

      Sorry, just had to stick this in. It's actually relevant from the market viewpoint. Big buyers are faced with a tough choice: Seagate produces lemons, Hitachi drives are expensive and unavailable, SSD's are eyewateringly expensive and have their teething problems. But unlike other two, SSD situation is rapidly improving, so it makes more sense to invest there.

      1. Anonymous Coward
        Anonymous Coward

        Re: A death and decline so easily forseen...

        This sea change has been obvious for the better part of a decade, it was not a surprise from left field like the iPhone. I cannot understand why Seagate ignored and continues to ignore it - I guess they are in denial like Joerg!

        They could have easily bought up one of the early flash SSD OEMs to get a leg up on the technology when their stock was worth more. They'd be in a far better position today, but it is too late now. They will either go bankrupt or be bought out, because they are too far behind now. While a market for hard drives will live on for years, they will be low margin capacity drives that will not support Seagate at its employee count.

      2. Alan Brown Silver badge

        Re: A death and decline so easily forseen...

        "Seagate and WD produce lemons"

        There, FTFY. That's as applicable to enterprise drives as to consumer ones.

        As a (small) enterprise buyer, I'm not going to be putting SSDs from WD or Seagate in my arrays (or any from their subsidiaries). I only bought Hitachi because they were a functionally separate unit from WD and now that's being folded into the mothership it's likely that quality will take a nosedive.

        Sammy PM863/SM863 drives are nice - and compared to enterprise spinners not terribly expensive. They draw far less power and the IOPs more than make up the price premium. The sad part is that most arraymakers won't use them as they'll just pack bare flash into a chassis instead.

    4. e_is_real_i_isnt

      Re: A death and decline so easily forseen...

      The problem is that the people actually designing and fabricating hard drives are not going to become SSD developers and technicians. All that would happen with the transition from one to the other is to keep the company name and a few top execs; all the rest will be discarded anyway.

      In Kodak's case, whether the name is still on the stock exchange only matters to shareholders and some retirees. The makers of their film products were never going to be fabricating image sensors and circuit boards.

      1. Anonymous Coward
        Anonymous Coward

        Re: A death and decline so easily forseen...

        The problem is that the people actually designing and fabricating hard drives are not going to become SSD developers and technicians

        How is that any different than the fate of those who designed horse drawn buggies a century ago, or who designed calculators 30 years ago? These are engineers, they should be smart enough to adapt to a new field (and should have seen the writing on the wall been planning this transition the last few years) They aren't unskilled laborers like assembly line workers or miners who may never be able to find another job that pays much beyond minimum wage when the coal mine closes or the manufacturing work is shipped overseas.

        1. Gene Cash Silver badge

          Re: A death and decline so easily forseen...

          > smart enough to adapt to a new field

          I'm an engineer, but I sure as hell couldn't go from writing Java to designing hard drive motors. I'm sure that road is equally difficult both directions. I *might* help design a SSD controller, but I can't see a company paying me big bucks for my amateur-hour efforts.

          1. Anonymous Coward
            Anonymous Coward

            Re: A death and decline so easily forseen...

            Engineers formerly writing firmware for hard drives could write firmware for other embedded devices, the skills are transferable. Engineers formerly designing the mechanics of hard drives might design other high precision devices like high end 3D printers. No one is suggesting a mechanical engineer is going to start writing software, or vice versa. What they at Seagate or WD is not so specialized that all their software or mechanical design talents are useless on other products.

            Anyway, what are the downvoters recommending as a solution? Ban or tax SSDs to keep the poor hard drive engineers employed? I doubt they had much sympathy when autoworkers lost their jobs to automation and probably said "hey technology advances, you gotta keep your skills current". But when it is engineers who are being displaced, suddenly it hits a bit too close to home for some, I guess.

    5. jake Silver badge

      @Steven Jones (was:Re: A death and decline so easily forseen...)

      "HDDs were simply never going to get remotely near SSD latency figures, and reduced latency is possibly the key performance issue in virtually every area of IT, whether it's processors, memory, storage or communications."

      When I was at Stanford, early one Saturday morning a Grad student drove to Berkeley on his motorcycle & came back with tapes of the over-night build of 3BSD. Our Professor, visiting from DARPA for a couple weeks/months (a dude by the name of Cerf, you may have heard of him), wondered how the hell our VAX had the latest version of BSD already running (10AM-ish), when the Switched56 connected source code system hadn't completed the download of the source, much less started to compile it.

      Biker's answer: "My motorcycle's latency might be sub-par, but it still has a much higher bandwidth capability than your network!". Cerf's reply? "Nice hack!" (Note that a variation of this quote is often attributed to Tanenbaum in 1996, but it was a fairly common meme around 1980.)

      Just something to think about ;-)

      1. Steven Jones

        Re: @Steven Jones (was:A death and decline so easily forseen...)

        Yes, yes we've all heard the story that the highest bandwidth is a truckload of tapes going down the highway. However, bandwidth is not the defining difference between SSD and HDD, it's latency and it's that which is insoluble.

        You can overcome bandwidth issues relatively easily (it's just expensive), but you can't beat the latency issue - and I speak as somebody who had to explain to a bunch of application developers on distributed systems that what they were up against was essentially the speed of light. We also know what root SUN went with their threaded architecture processors which were pretty good on throughput, but crap on latency which made them highly unsuited to some problems. Latency matters, and it matters more and more as systems get integrated yet we sill demand instance responses.

  2. thexfile

    Mechanical drives are becoming a thing of the past... Long Live King SSD!

    1. Voland's right hand Silver badge

      For day to day use - yes. Media - NO

      DVDs and Photos rack up capacity at a rate where only spinning rust is financially affordable in consumer space. However, the overall tendency for that is P2P to decline and streaming services to grow. So even that is not a long term growth driver.

  3. picturethis
    Mushroom

    Statements sound similar to..

    the statements made in the mid-90's by Kodak CEO's regarding film vs. digital cameras:

    1) Film will always be around because it's higher "resolution" (better performance)

    2) Digital Cameras are too costly

    Right...

    Either Seagate had better change it's mix of offerings (higher % SSD, lower % HDD) or it will go the way of the dinosaur. I'm betting on the stubborness (or lack of reality) of Seagate CEO's. Watch for the golden parachutes to be packed, ready for deployment within 5-7 years...

    If you work for Seagate, start looking for a new job now while you have time, it doesn't look good longer term. (You'll be competing with the Intel crowd now though.. best of luck in these tough times).

    "It's a recession when your neighbor/friend loses their job, It's a depression when you lose your job".

  4. another_vulture

    Enterprise : capacity vs performance

    This article at least distinguishes the two types of "enterprise" drives: "capacity" and "performance". It's clear that "performance" HDDs are already completely obsolete due to SSD performance. It's less clear that "capacity" HDD is obsolete, as they still have a large 15x advantage in $/TB. I do wonder however if there is any real advantage in a "enterprise capacity" HDD versus a cheaper "commodity capacity" HDD, given that performance (IOPS) ceases to be an interesting metric for this application.

    1. Anonymous Coward
      Anonymous Coward

      Re: Enterprise : capacity vs performance

      Seagate wants there to be a difference, but there is not. Since enterprise capacity and commodity capacity drives are both SATA, these are the only differences:

      1) enterprise drives are less dense, to achieve a lower undetected bit error rate (and/or use a larger CRC than commodity drives do)

      2) because of the way they calculate MTBF, a lower undetected bit error rate translates into a higher MTBF (not that anyone really looks at those anymore, but people expect enterprise drives to have a better MTBF)

      3) because they are less dense, they cost more per TB

      4) they may undergo some additional testing cycles, which makes them cost even more and be further behind the comparable commodity capacity drive at introduction

      None of these matter except #1, and you can mitigate the issue - i.e. if your upper layers use CRC checks for data integrity, you can exceed the enterprise drive's undetected bit error rate with a commodity drive by as much as you feel you need. Or if you are EMC, you modify the firmware and format with larger sectors so the array can check the CRC in a second layer check and reduce that undetected bit error rate to ridiculously low levels. MTBF doesn't matter, especially if you are getting more TB for the same money, and vendors who care about extra testing (i.e. EMC and friends) already do that themselves.

      1. Alan Brown Silver badge

        Re: Enterprise : capacity vs performance

        > 2) because of the way they calculate MTBF, a lower undetected bit error rate translates into a higher MTBF (not that anyone really looks at those anymore, but people expect enterprise drives to have a better MTBF)

        However... they don't - and haven't for at least 15 years.

        The failure rate of certain types of Seagate Enterprise "storage" drives (constellations - 250% during the 3 year warranty period) is only exceeded by the failure rate of the more lemonny consumer drives (Barracuda DM and DL series - 180% during the 12-month warranty period)

        Interestingly, one of the array suppliers who fitted constellations on initial supply (SGI) are now replacing them with WD Black (not Red or Red+. Go figure)

        1. Pascal Monett Silver badge
          Trollface

          Re: WD Black

          That is indeed curious. On the other hand, I'm an obedient little cheese-eating surrender monkey ; I set up my Synology NAS with 4 3TB WD Red HDDs (well, there's a Seagate one in there just for kicks).

          I check out what WD Black is and find something that really makes me laugh : WD is making HDDs for, get this : CLOUD STORAGE !

          If ever there was a "sucker born every minute" product, this has to be it.

    2. Alan Brown Silver badge

      Re: Enterprise : capacity vs performance

      "It's less clear that "capacity" HDD is obsolete, as they still have a large 15x advantage in $/TB"

      You're comparing consumer capacity drives with capacity enterprise SSDs

      The price difference between capacity enterprise drives and capacity enterprise SSDs is only a factor of 3-5 and that gap keeps narrowing every 6 months.

      Regarding IOPS, it _does_ matter inasmuch as a capacity array with sufficient performance makes a performance array unncessary - or if still needed, _much_ smaller.

  5. David Halko

    Move to Capacity with Innovation

    Rotating Rust manufacturers must find a way to innovate out of this problem.

    - Possibly push technology from IOPS (losing war with SSD) into Bandwidth & Capacity (to eat market share from Tape) may guarantee additional years.

    - A marketing push for MAID with SSD cache for backup purposes may be a great market to push towards, giving Rotating Rust a practical engineering effort to invest into SSD technologies.

    - As odd as it may sound, perhaps engineering larger diameter (but lighter) platters could increase capacity substantially to compete cost effectively with SSD manufacturers [building $1B fabs every time they need to shrink.] Once again, not to replace SSD's, but leveraging SSD's for cache.

    - Also, perhaps a focus on long term storage: WORM for 10-20 years. This could give Rotating Rust a way to chew up ladder to supply solutions that traditionally higher level storage providers handle today.

    - With The Internet expanding ever so quickly, it must be archived somehow, it will be archived, and government agencies are able to pay a premium since they merely tax more to get their income. With the explosion of IoT, there will be more traffic to be captured than ever before. Seldom accessed, forever stored.

    - Integrated storage hierarchy systems, bundling higher amounts of RAM, Super-Capacitors, Batteries, Flash, Rotating Rust in the same case in a modular way. Target long-term storage, again. Perhaps with hot-swappable sub-components (Some efforts being seen on including batteries in consumer items may be a first-step in a long term effort.)

    - Push towards higher availability solutions for enterprise servers (i.e. press a button in the front of an integrated Rotating Rust drive to temporarily disable flash to perform a hot-swap in the hard drive chassis, similar hot swap of a thin/long integrated battery that could be removed hot.) Remove innovation support requirements from server OEM's or OS vendors.

    Rotating Rust has opportunities, but investment & diversification may prove to be challenging and disrupting to people above & below them in the storage hierarchy.

  6. Anonymous Coward
    Anonymous Coward

    They need to act, it may already be too late

    I have often wondered why Seagate has not acted sooner. SSDs were coming on strong four years ago.

    15K SAS drives are dead, 10K SAS drives are dying. Only things left will be the slow 7200RPM capacity drives. For enterprises using SSDs with inline deduplication and compression, and the huge power consumption savings have actually rendered many SSD solutions less expensive to own over a 3-5 year period. And SSDs are far more reliable and have lower UBER rates than HDDs.

    In the PC and laptop space the better performance and lower power consumption favor SSDs and the cost differential is minor. On laptops you will get longer battery life.

    And there is more coming. 3D TLC NAND and eventually QLC will begin killing off the capacity HDD segment although this will take longer.

    WD saw this coming an that's why they have been making investments. They bought Velobit (caching SW), Virident, STEC, Skyera, and now SanDisk. Seagate has done practically nothing although they have a good SSD controller (SandForce) which they bought from LSI/Avago. Seagate needs to acquire a NAND fab (by buying SK Hynix), integrate the chips with the SandForce controller and aggressively develop SSDs, or just die a not so slow death.

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon

Other stories you might like