back to article Disk death: Three-quarters of PCs will run SSDs by 2020

Total disk drive shipments are going to plummet by 2020, with raw SSD cost getting cheaper than disk and SSDs taking over from disk in notebooks. Analyst haus Stifel Nicolaus's MD, Aaron Rakers, has taken a close look at Gartner’s HDD and SSD projections over the next few years. He writes: “Gartner is currently forecasting …

  1. Blergh

    Disk death: Three-quarters of PCs will run SSDs by 2020

    I was going to say well duh that is so far away I'd be surprised if there was any disk left. But then I realised it is only 4 years away and not the 14+ years my initial assumption thought it was.

    When did 2020 stop being some mythical time so far into the future that it might never happen?

    1. Jeffrey Nonken

      Re: Disk death: Three-quarters of PCs will run SSDs by 2020

      I remember when 2000 was that, and many excellent science fiction stories are now badly dated because of it. So a looming 2020 is less of a culture shock for me. :)

      1. David 132 Silver badge

        Re: Disk death: Three-quarters of PCs will run SSDs by 2020

        I remember when 2000 was that,

        You, me, and all the other Squaxx dek Thargo.

  2. Alan Brown Silver badge

    only 3/4?

    The only thing preventing SSDs taking over _now_ is that supply can't keep up with demand.

    Anything smaller than 1TB is pretty much 100% SSD already.

    1. banalyzer

      Re: only 3/4?

      That depends, supply is restricted because many manufacturers were unsure about the value of investing that heavily. If the price starts to drop too dramatically then the cost of investment may not be recovered. Although some have announced new fabs things like xpoint etc are putting a brake on that.

      I updated my laptop from an 750 GB SSHD with an 8GB Flash cache to a 500GB SSD earlier this year.

      The boot time is directly comparable to the extent that I haven't actually noticed any real difference between them. As far as loading of programs go, the difference is easier to discern the larger the program is but is still not onerous.

      When I'm given a laptop to repair, most still have a standard HDD with only RAM cache.

      I suspect that the life of HDD could be extended by sticking a small SSD in front thereby giving the perception of being much faster because the boot time is comparable and this is what most people consider.

      Once the supply ramps up properly and the cost of a 1TB SSD is less than twice the cost of a 1TB HDD then the consumer market will collapse

    2. Pirate Dave Silver badge
      Pirate

      Re: only 3/4?

      "Anything smaller than 1TB is pretty much 100% SSD already."

      Not for those of us still using enterprise SAS drives - 300 and 600 gig spinning drives are still bread and butter for some of us. Not all of us have the data growth needs of Facebook or Google (or their budgets).

  3. AndrewDu

    Well I hope they're going to improve the reliability a bit.

    I built a mini-server to use as a Domain Controller for the home network; for fun I built it with no moving parts at all, so of course it had an SSD.

    After replacing the damn thing three times in the first year (and we all know what fun it is when a Domain Controller croaks, don't we?) I changed it for a Samsung 2.5" laptop drive, which has been running without a beat dropped for about three years now.

    Of course I realise Enterprise SSD's are better than this, but have you seen the prices?

    1. David 138

      I'm sure they were saying there is very little in it between enterprise and consumer reliability. SSDs are alot better than they used to be and i cant stand a using a computer without one. If you buy a computer nowadays without one you should be shot.

    2. Gerhard Mack

      If a single drive failure takes out your domain controller, you are doing it wrong. As an aside, I lost 3 spinning hard drives in two years on a two drive RAID server so spinning is not much more reliable.

      1. Pirate Dave Silver badge
        Pirate

        "lost 3 spinning hard drives in two years on a two drive RAID server so spinning is not much more reliable."

        We're an HP Proliant shop here, and I have to say the 2.5" SAS drives HP has been using the past 5-7 years or so are pants. About 25% of them seem to fail within the first 2 years (although so far the replacements seem to be holding up well). I contrast this to the 3.5" U320 drives I bought from IBM 10-12 years ago (for their xSeries servers) which had a less than 10% failure rate (and most are still running - but not doing anything "important" since they're so old).

      2. John Tserkezis

        "As an aside, I lost 3 spinning hard drives in two years on a two drive RAID server so spinning is not much more reliable."

        Yeah, so did I, till I realised that I shouldn't have been using desktop drives in an array environment. Now, with RAID or Enterprise grade drives, they're still going fine.

        You can't just throw any old drives in your array and magically expect them to last.

      3. AndrewDu

        Fair comment.

        Of course I should have been using RAID etc etc we all know that.

        Perhaps I didn't make clear this was a DC on a home domain, not real live work in actual production.

        But cleaning up the DNS was still a messy business.

    3. Alan Brown Silver badge

      "so of course it had an SSD. After replacing the damn thing three times in the first year "

      What model was it? OCZ have been spectacularly unreliable as a f'instance.

      "Of course I realise Enterprise SSD's are better than this, but have you seen the prices?"

      Samsung SM863s are about 30% more than the 850Evo versions (or about 15% more than the 850Pro)

      And as has already been pointed out, if losing a single drive knocks out a server, you're doing it wrong.

      1. Anonymous Coward
        Anonymous Coward

        OCZ ok (so far)

        Have an OCZ from Jan 2013 and for most of the time have been running Windows Media Centre with the pause live tv feature enabled.

  4. Doctor Syntax Silver badge

    2020 is 4 years away. Can we have some examples of what these same analysts were saying about 2016 back in 2012? It's absolutely no use quoting this stuff without some way of judging the sources' reliability? It should be standard procedure.

    1. Flocke Kroes Silver badge

      I think it is in the article

      Look at the first graph (revenue), along the bottom. The first four years are do not have an E, implying that the blue rectangles are actual correct numbers arrived at by having someone follow every single disk out of the factories to the customers and getting the bank records of the customers and the manufacturers to find out precisely how much money changed hands. The blue rectangles for years 2016E onwards are clearly marked as guEsses from Stifel.

      The blobs with lines are all Gartner estimates, but each line came from a Gartner estimate published at a different date. This clearly shows the Gartner estimates are 'correct' for the date published, but no use when looking for future perfomance.On the other hand, Stifel do not have historical predictions on the graph at all.

      The second graph (capacity) works precisely the same way, but the third is complete Gartner. If you predict HDDs revenue continues to fall (instead of Gartner's near constant prediction) and you go for modest increases in capacity (instead of Gartner's wild optimism) you get a vaguely falling cost/GB for HDDs (using Gartner predictions you should have a rapidly falling line instead of the level one on the graph).

      I will leave my predictions for the future until later, just like Gartner would if they wanted to be accurate.

    2. Anonymous Coward
      Anonymous Coward

      Gartner?

      You are wise to check the quality of the source material and their track record of accuracy.

      This is the company after all, that predicted that by the mid 2010's, fibre channel would go the way of the Dodo and we'd all be running our kit on FCoE and iSCSI.

      1. Gerhard Mack

        Re: Gartner?

        They also predicted that the Itanium would rule the server market.

        1. JC_

          Re: Gartner?

          I'm reading this thread on those Windows Phones we all have now.

    3. John Brown (no body) Silver badge
      Thumb Up

      "2020 is 4 years away."

      Exactly! Most of the kit going out of our doors is customers still speccing spinning rust PCs and most it is going with 4 or 5 years on-site cover. They can choose SSD if they want, but most are buying to price, nor performance. The tipping point may well come in 2020, but it won't be 75% already on SSD by then. It might be the point where SSD overtakes spinning rust in sales but not in installed userbase. Maybe.

  5. Efros

    All of mine

    have at least a boot and OS SSD. It is the single best upgrade you can give a PC. The only machine I have that has a substantial amount of SSD storage is the games machine which has a couple of 480GB SSDs in RAID 0. Does my Skyrim installation no end of good.

  6. theblackhand

    So...

    Gartner have been reaching into the magic butt and pulling out the contents again. Think they are a little optimistic within 5 years although they will probably be correct within 10 years.

    Projections for the global PC market by 2020 put it at around 290 million with a breakdown of around 170m laptops and 120m desktops. I'd expect close to 100% of laptops to have SSD's by 2020 (currently ~33% and prices for laptop SSD's/HDD's are nearing parity so no real reason for not moving to HDD's other than high end capacity).

    On the desktop side, will they reach around 20% selling with SSD's? Assuming the desktop market has drifted even further to value systems (i.e. around 47% value, 47% mainstream and 6% performance in 2012), they would need around 100% of performance desktops and 25% of mainstream PC's to include and a likely bump in price. Sounds a little optimistic.

    TL;DR: Gartner produces report showing next PC boom for hardware manufacturers ~10 years after PC hardware booms ended

  7. Anonymous Coward
    Anonymous Coward

    The next thing

    those cheeky anal (ysts) will be telling us will be the Death of Tape backup.

  8. Mingy

    Why does anybody pay for Gartner research? If they had to dramatically revise their HDD forecast 7 times over the past few years it's pretty clear they have no understanding of the HDD industry.

  9. Anonymous Coward
    Anonymous Coward

    Death - A Tiny Exaggeration

    Actually a fairly big exaggeration - "revenue is expected to be down two per cent over this timeframe", so not dead.

    OK, HDDs are hitting a mature phase, but like tape expect a long tail. Death is just a headline grabber and not what we will see in the next 5 years. Primary data is likely to mostly move to flash over this period but HDDs will not be disappearing (other than from laptops and to a degree in their 15K/10K flavour). They are just moving down market. Not exciting or glamorous but still many billions of dollars. Not dead, just old.

    1. John Brown (no body) Silver badge

      Re: Death - A Tiny Exaggeration

      The death of the floppy was predicted when Apple started shipping iMac without a floppy drive.

      The death of the CD/DVD drive was predicted when Apple started shipping PC/Laptops without optical drives.

      Neither are completely gone yet, optical drives still having a respectably long future still to come IMO.

      1. Anonymous Coward
        Angel

        Re: Death - A Tiny Exaggeration

        and of course Vinyl totally disappeared after CD's appeared, and CD's after MP3 etc

  10. ben_myers

    For once, I might even agree with Gartner

    Having replaced failed no-longer-spinning hard drives in three laptops in the last few days, I agree that the world needs to move full speed ahead to SSDs. Now let's get the SS price points down!

  11. quxinot

    "....with raw SSD cost getting cheaper than disk..."

    I can't wait for 1Tb SSDs to cost $40, but unless something drastic happens, that's not going to be in the next four years.

  12. Evito

    FLASH FORWARD conference on flash and SSD, featuring Chris Mellor

    A one-day conference focussed on flash and SSD storage and aimed at end-users involved in deploying and managing IT storage will be taking place on 14th June in London. The Register's Chris Mellor will be discussing 'Future directions and impacts on datacentres and application time to market' at the event.

    Agenda and further info at http://flashforward.io.

    Disclaimer: Evito Ltd is the organiser of FLASH FORWARD

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