back to article IBM will soon become sole gatekeepers to the realm of tape – report

Spectra Logic's Digital Data Storage Outlook 2017 report predicts IBM will emerge as the sole tape drive manufacturer. The 24-page report is available for download (PDF) with no registration. This is Spectra Logic's second such annual report. It reviews the rise of flash and its effect on disk, tape and optical storage media …

  1. Alister

    Spectra Logic's Digital Data Storage Outlook 2017 report predicts IBM will emerge as the sole tape drive manufacturer.

    Looking at the way IBM is rapidly disappearing up its own backside, this may be wishful thinking!

    It ( / they) seem to be divesting themselves of anything hardware related, in an effort to become all cloudy and edgy, and service oriented.

    1. deevee

      have to agree, IBM will soon be purchased by Apple or google and disappear into the history books as yet another enterprise that thought selling "fluff" (i.e. anything that costs nothing to produce) and offshoring anything physical to the cheapest, dumbest, illiterate monkey they can find in India, was the "way of the future"

      I'd be betting on Oracle/Storagetek for a future in tapes.

      1. returnofthemus

        I'd be betting on Oracle/Storagetek for a future in tapes.

        LMAO!

        This public cloud wannbe is still fighting for it's future in database, let alone being able to INNOVATE in Tape Technology, which is why they're giving away their hardware at maintenance renewal time.

      2. PetF

        Oracle did stop the development of the T10000e several month ago.

        I will not bet a copper on Oracle to keep tape alive.

  2. Peter2 Silver badge

    Another, and perhaps more optimistic, conclusion is this: "Spectra believes that a probable long-term scenario is one in which flash technology and tape will coexist, and become the prevailing storage technologies for online and archive needs, respectively."

    I think that's reasonably probable.

    Though if an HDD manufacturer axed R&D and spending on new production equipment and just kept churning out 1TB drives on their existing equipment and got the price down to something like £20 per terabyte drive then that'd screw SSD's on storage and also be in a position to screw the rationale for tape- at that price if somebody created a briefcase full of HDD's connectable via USB3 you'd get the same storage as LTO7 at a similar cost without needing an expensive tape drive.

    Difficult to foresee, the future is.

    1. Anonymous Coward
      Anonymous Coward

      What would be the point in churning out cheap 1 TB drives versus more expensive bigger drives that have a lower cost per byte? There's no market for cheap 1 TB drives, other than for laptops/desktops selling at the bottom of the market. It wouldn't "screw" SSDs because that's a small percentage of the overall NAND market.

      Hard drive OEMs could drop R&D and concentrate on getting the price down for really big 16TB or whatever the max is. That's what cloud providers want, and it will be years before 16TB SSDs can cost less than what they could sell a 16TB hard drive for - especially if the plant has been paid for, R&D have been laid off, etc. five years from now.

      1. Alister

        What would be the point in churning out cheap 1 TB drives versus more expensive bigger drives that have a lower cost per byte?

        It's an interesting question, at the end of the day spinning rust will break, and I definitely would not be comfortable in having 16TB of data on one drive in an array, the rebuild time would be horrendous.

        I for one would rather have larger arrays of smaller drives (although I'm not sure that 1TB isn't too small nowadays as you say.

        But there must be a sweet spot where price-per-byte and the time taken to recover from a failure intersect, although I haven't looked into it, I would guess around 4TB.

        1. Anonymous Coward
          Anonymous Coward

          More drives wastes slots in an array. You'd be better off doing dual parity like RAID6 14+2 so the rebuild time doesn't matter (unless you really think a triple failure is likely during the rebuild window)

          1. Peter2 Silver badge

            The exact size of the HDD doesn't really matter. My premise was that a company could go down a "no further R&D, ruthlessly reduce production costs" route and significantly impact the market, potentially keeping HDD's alive more or less indefinitely and potentially even wipe out tape for off site archiving.

            I don't honestly think it's likely since tape does have a lot of compelling advantages but it's far from being impossible that radical action could keep HDD's in more or less mainstream use for a lot longer.

          2. Alan Brown Silver badge

            At 10TB+, the chance of a triple failure is about 1-2%.

            And all those drives have about the same uptime.

            1. Anonymous Coward
              Anonymous Coward

              Not sure where you got that 1-2% triple failure stat given that drives typically fail at a 1-2% annual rate, but higher order RAID with 3 or 4 parity drives is still a better/cheaper solution for cloud capacity for colder data than having more arrays to provide the necessary number of slots.

              1. Alister

                It's taken me a while to respond but I wanted to check my facts before posting.

                I remembered reading an article which stated that as larger and larger capacity disks are used in RAID arrays, the more likely you are to see non-recoverable read errors on some sector of the disc. If you take the manufacturer's stated figures, the incidence of URE is about 10^14.

                Which means that once every 100,000,000,000,000 bits, the disk will be unable to read a sector. That figure of 10^ 14 is roughly equivalent to 12TB. So if you use drives of 12TB or more, then you will have a URE at some point.

                So imagine if you have a disc failure in a RAID array using 12TB disks. The time taken to rebuild the array means that there is a significant chance of a non-recoverable read error on one of the other drives during the rebuild (BER / URE). That would be GAME OVER.

                So using more, smaller drives in larger arrays in RAID 6 or better, greatly reduces the chances of a multiple failure and data loss.

        2. phuzz Silver badge

          "at the end of the day spinning rust will break"

          SSDs and tape aren't 100% reliable either, which is why RAID (to improve uptime), and backups are just as important as they've always been.

          Last time I checked (a few months ago), the price/capacity sweet spot for consumers was 3TB drives, unless you can find a good deal on a 4TB one (it's sometimes cheaper to pick up an external drive and pull it out of it's case).

      2. Anonymous Coward
        Anonymous Coward

        Magnetic Drives?

        I'm curious if you honestly feel like the future of backup destinations is in magnetic tape media. Although SSD prices are still higher, the prices are dropping very quickly and reliability make recovering from such media dependable. As sitting spinning drive will propose problems if all of a sudden it needs to produce data stored after an extended amount of time (remind you of something)?

        1. Peter2 Silver badge

          Re: Magnetic Drives?

          I'm curious if you honestly feel like the future of backup destinations is in magnetic tape media. Although SSD prices are still higher, the prices are dropping very quickly

          Honestly, I think the future of offline offsite backups is in tape and I think HDD's are toast except possibly as online onsite single backup copies to save the hassle of getting last nights tape. SSD's aren't going to be affordable enough for offline offsite backups for a LONG time.

          The catalogue price for a single LTO6 tape is £30 for 3.25TB uncompressed capacity. All you need to is ask, and you can get about 20% off of that if you have an account manager and are buying any sensible number, but anyway.

          SSD storage costs are not in that ballpark. To get to 3250GB you'd need an awful lot of individual drives. Basically your looking at about £1000 worth of SSD's for the same level of storage and most companies doing tape run a month series of tapes.

          Yeah, the price of flash is falling and that's a lazy comparison of looking at a last generation tape (LTO6 is last gen, LTO7 is available with double the storage) versus multiplying the space of the cheapest consumer SSD's which you wouldn't do, but the prices are roughly indicative. Even if the price fell by a factor of ten, tape vendors would still be saying "SSD's are triple the price!"

          And that ignores that there is a tape roadmap out to LTO10, with a native capacity of 48TB per tape which the tape vendors will be marching along while the SSD price falls which is probably going to widen the gap to continue making tape viable.

          So for offsite offline backups, there is little competition. Online backups are another matter, but for the cost versus risk factor it's not worth not doing offline storage of company endingly critical data.

    2. returnofthemus

      I think that's reasonably probable.

      Already being delivered Today TODAY!

      Welcome to the Cognitive Era ;-)

  3. Anonymous Coward
    Anonymous Coward

    Is that headline correct?

    Headline: IBM will soon become sole gatekeepers to the realm of tape – report

    "Further: "Cloud providers will mostly adopt LTO, and given their strength in purchasing overall tape technology, this will lead to a greater percentage of LTO versus enterprise tape technology. Enterprise tape, with two suppliers, appears to be an over-served market and Spectra predicts that, at some point, the market will converge to one."

    It is Spectra's opinion that IBM will be the sole manufacturer of enterprise-class tape drives and media in the years to come."

    I read this as saying IBM will become the only ENTERPRISE tape vendor, but the majority of the tape market will be LTO based. Given HP is currently the biggest LTO vendor and enterprise tape makes up around 10% of the total drive market and shrinking (90% LTO), my vaguely educated guess is that IBM has 30% of the overall tape drive market and I would be surprised if the other three vendors planned to just walk away...

    1. returnofthemus

      Re: Is that headline correct?

      I think you read correctly, it's not a case of the other three vendors walking away, it's about them being able to step up, to me it doesn't look like any of them would have the apetite, let alone the R&D capability.

  4. earl grey
    Unhappy

    I miss tape for my PC

    Yes, tape. And no, I'm not trolling.

    1. bombastic bob Silver badge
      Devil

      Re: I miss tape for my PC

      I still have an old QIC/Travan drive banging about in the boneyard, and a couple of tapes to go with it. They store >1Gb per tape. Wow!

      https://www.cnet.com/products/hp-colorado-t3000-tape-drive-travan-1mbps-floppy-series/specs/

      and I have a box of IOMega 'zip' drive cartridges, too. 10Mb each!

      Amazing the money we all spent on that kind of tech, back in the day...

      [it was SO quickly obsoleted]

      1. Flakk
        Trollface

        Re: I miss tape for my PC

        I have a box of IOMega 'zip' drive cartridges, too.

        I bet on a quiet night you can hear them going "click click click click click".

      2. yet_another_wumpus

        Re: I miss tape for my PC

        Zips were 100MB. I had 200MB in my laptop harddrive, so I bought the first one I saw. I then bought one for my father when his birthday/father's day/Christmas came around.

        The 1G tape took a bit longer, and the CD-R obsoleted it faster than I expected. It also remains an example of "things I had linux drivers for but never windows drivers (the driver cost roughly as much as the tape drive)" [the counter example was my first DSL modem. And I really wanted that on my Linux box, windows just wasn't up to the internet back then].

  5. John 104

    For 2.5mm spinners to completely go away, SSDs need to become very cheap in the enterprise. And I mean cheap. We are coming up on end of life on some of our first gen SSD deployments. Its not pretty. SSDs don't gradually tip over like a traditional hard disk might. They just stop working. Placed in a RAID, you can have sudden, dramatic loss of RIAD and no rebuild option. Better hope you have good backups at that point. (we just moved to LTO7 and I'm quite happy with the capacity).

    1. Alan Brown Silver badge

      "SSDs don't gradually tip over like a traditional hard disk might. They just stop working. "

      They tend to give plenty of warning IF you bother to pay attention to the SMART parameters.

  6. Herby

    We've come a long way...

    From 10.5 inch reels of tape that stored 24Mbytes at 800 bpi.

    Then again, it did have a certain romance to it as you threaded the drive (the ones I used had the source reel on the left side!).

  7. Anonymous Coward
    Anonymous Coward

    Good riddance HP

    After fucking over clients - including renegging on the "lifetime warranty" on media.

  8. Anonymous Coward
    Anonymous Coward

    I love that graphic! I recognized it right away -- my god, I've lost count about how many times my friends and I had tried to make it through that DnD module. I had a chuckle at my desk when it was used in the context of tape... especially when I remember wrestling Spectralogic Bullfrogs with AIT tapes and whole slew of other nastiness early in my career. Carry on El Reg, carry on... (=

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