back to article IBM wheels out bleedin' big 15TB tape drive

IBM has brought out a TS1155 tape drive as an update on the existing TS1150, offering 15TB raw capacity, half as much again. These are proprietary IBM format tape drives. For comparison the open standard LTO-7 format offers 6TB raw capacity (15TB compressed at 2.5:1), with the coming LTO-8 reaching 12TB raw, well below IBM …

  1. kain preacher

    I'm not sure I want to be the one to have to a do a restore with size tape. I see two things happening. it will take for ever or during the middle of the restore it barfs.

    1. Dwarf

      @kain

      How is that any different from any other type of tape drive in existence ?

      1. Steve Davies 3 Silver badge
        Paris Hilton

        Shudder

        I still recall the issues we had with QIC-25's around 1978. The tapes were good for about 4 complete operations before they needed to be junked. One system booted RSX-11S from them. That was good for about 10 reboots before barfing.

        I doubt even Paris barfed that many times

    2. Anonymous Coward
      Anonymous Coward

      "it will take for ever or during the middle of the restore it barfs"

      That's how they get you, see?! That wouldn't happen as often if you bothered to license and enable recursive accumulating backhitchless flush with byte-bowl file freshness forward-winding virtual tape quantum blockchain cloud VR AI and the accompanying wrap-around headset with head's up display! Stop the maddness! And also scrub your read heads with steel-wool or an old candy cane left over from the holidays, that should cure your tape ills. An no need to mark your tapes, the tape robot will figure it out from the tape leader. Backups are a trifle, mates.

      *continues marking the edges of old CD-ROMs with a blue Sharpie*

      Seriously, I once had another group (windows admins *shakes head*) visit my tape library and "borrow" tapes they "thought" were not in the use for their next backup run... only they didn't ask my library about the tapes, they had been "sharing" them between two unconnected, but same vendor, backup systems, and thought nothing of it. That stopped the day I found out about this odd practice. Weirdos.

  2. Alistair
    Windows

    And they still will go to hell in 6 years....

    and not be readable....

    1. Anonymous Coward
      Anonymous Coward

      Re: And they still will go to hell in 6 years....

      Do you realize how much money the IT manager saves by purchasing the lowest price tapes available?! Come on, man! Nobody is going to request a restore from it anyway, so no harm done, yes? ;)

      Data, if you compress it down to a single bit, will anyone notice? And will it be a one(1) or a zero(2)?

  3. Anonymous Coward
    Anonymous Coward

    But, but tape died 20 years ago. Lots of people said so.

    1. Storage Girl

      Storage Girl

      Yes, many said so. The ones who did not sell tape solutions.

      Some reasons that tape has a bright future are:

      Data generated in the internet of things is outstripping all available disk and flash and tape supply.

      Tape is the least expensive storage technology by a mile.

      Tape has linear scalability in areal density growth for at least a decade.

      Think about this. Flash has already passed disk in lower TCO than disk. If I were selling only disk I'd be very nervous about the future. Flash provides greater performance at lower TCO than disk and Tape provides lowest TCO overall. The only reason to buy disk is the first point above.

  4. Anonymous Coward
    Anonymous Coward

    So, IBM created a Tape Drive. Why is this news? This isn't going to stop the Financial Bleeding at IBM and is far less interesting than their Contractor Hiring Freeze and 20 quarters of consistent financial losses. Don't expect everyone to run to IBM for this Tape Drive or the even more pointless WATSON, that was recently referred to as a "Joke", by an Investor being interviewed on CNBC.

    1. the spectacularly refined chap

      Yes, over those twenty quarters the losses are very consistent. As in:

      • Made shit loads of profit
      • Made shit loads of profit
      • Made shit loads of profit
      • Made shit loads of profit
      etc etc.

      So you don't really care about the drive? Of course not - you'd much rather miss the point of an entire series of articles and bash on about something you know nothing about.

      Yes, the drive itself is mildly interesting if you want that sort of thing, but personally I'd still tend to stick with LTO for those things not on D2D.

      As a side point - when did ordered lists disappear from the allowed HTML syntax?

      1. Anonymous Coward
        Anonymous Coward

        You're wrong. IBM isn't and hasn't made any "Profits" in 20 quarters. That doesn't come from reading news articles about it or spouting un-informed opinions. That comes from WORKING THERE and being among contractors laid off during the sudden Contractor Hiring Freeze. Read that article on this site, then try to defend IBM again.

        1. the spectacularly refined chap

          Cuts != losses. Declining revenues != losses. LOOK at the figures, e.g. https://www.ibm.com/investor/events/earnings/1q17.html A PROFIT of $2.3bn in the first quarter. It is the same across the board.

          What do you do there? The lady or the post boy?

    2. Anonymous Coward
      Anonymous Coward

      Phew, this news item was almost bordering on the positive for IBM on the Register, and I needed my daily dose of IBM bashing. Thank you!

  5. Kevin McMurtrie Silver badge

    First, let me say that I'm very impressed with being able to manage 360 Mbps on a ribbon of tape flying by despite mechanical imperfections on large components, thermal distortion, dust, and even air itself getting in the way. I have destroyed many tapes just trying to recalibrate an old analog VCR.

    On the other hand, 15 TB is not impressing me. These cartridges may be cheaper than a hard drive but they're not much smaller and they need incredibly specialized mechanical support. Customer and IBM would have to agree to support each other for a long time to make this worth while.

  6. Christian Berger

    Why does nobody build disk libraries?

    I mean tape is only really economical when you have a library, so obviously there are tape libraries acting as a huge tape changer. Why don't do people do this with optical disks. When you only handle them with robots you can get around one of the main problems, scratches. Plus disk formats are rather stable so even in 20 years you'll probably still be able to get something that can read Bluray disks easily.

    1. Anonymous Coward
      Anonymous Coward

      Re: Why does nobody build disk libraries?

      "Plus disk formats are rather stable so even in 20 years you'll probably still be able to get something that can read Bluray disks easily."

      That's what I was hoping for when I transferred all of my Zip-disk archives to HD-DVD. Media longevity...

    2. J. Cook Silver badge
      Boffin

      Re: Why does nobody build disk libraries?

      They do make them; are your referring to something like this?

      https://kintronics.com/solutions/optical-jukeboxes-and-libraries/

      (first hit from google "optical jukebox" that wasn't a wikipedia link)

      TL;DR - Yes, but they have the same problems. They are a bit faster for access times.

      Optical jukeboxes tend to have two problems:

      1) Relatively low data density compared to tape. (a BD-RW gets you what? 50 GB tops per disc?) While the discs can be crammed into a smaller volume, You need more of them to equal tape. One unit is 5 feet by 15 inched by 28 inches and only manages ~70 TB max capacity. An equally sized tape library will beat that without breathing hard. (oddly enough, they cost the same...)

      2) They suffer the same problems as tape changers. In the ~8 years that I've been overseeing backups at my current employer, we've gone through four tape libraries; the first one was died from old age and 'we don't support that model anymore'-itis, second had the robotics die on it requiring a chassis replacement with unit #3, which was superseded by the fourth, which has had the controller and most of the drives on it replaced in it's 5 years of operating life. They both have a good deal of high precision moving parts which are fiddly, finicky, and sometimes downright cranky.

      They do have some pluses, though: faster access time than tape, and depending on the media fed into them, better archival longevity. They also have random access capability as well, so you don't have to de-spool 3/4 of the tape to get the single 100 MB file you are after near the end of the media.

      My first corporate job was with a credit union that had two optical libraries: a moderate sized one for mortgage records (about the size of a half-cabinet), and the second was this giant beast of a unit for check image storage. (it was larger than a double wardrobe) and that was in the late 90's, so that was CD-R or M-O media.

      1. Dwarf

        Re: Why does nobody build disk libraries?

        There was also the Plasmon UDO, archive platform, which looked really good from a reliability perspective.

        Last I heard they had gone, but it seems that the domain still exists, but the driver compatibility seems to be about 10 years old, so looks like its gone to the great big bit barn in the sky.

  7. John Smith 19 Gold badge
    Unhappy

    The ideal tool for TLA's looking to spy on everyone, all the time, forever.

    What did you think this is for?

    Sinister Orwellian uses aside you have a 15TB chunk you can just eject and when you want to double your capacity you buy another tape.

    Thing about tape. Zero power storage. No fear of bits zapped by cosmic rays. HDD has 100s of parts and 10 of 1000s of connections, nearly all single point failures.

    Yes tape drives are mechanically complex (but so were all designs of VCR's, and yet people managed to make those in large quantities to ever improving accuracy) but I've never understood why changers are so expensive other than "because we can charge that kind of money." I think there's a fascinating mechanical engineering research project to be had in adding a changer to a manual tape reader using 3d printing, because if you can meet the alignment specs once you publish the design anyone can do it.

  8. TWB

    Compression question

    Honest question so please be gentle.

    In the article it says compression can be used to up the capacity of the various tape systems - I presume this using lossless data reduction techniques - the equivalent of zipping up files on write and unzipping on read - correct me if I'm wrong.

    My question is - what happens if all the data is already compressed .e.g .jpgs, mpegs etc?? (Where I worked, we had an LTOx archive just for video - so no compression used)

    Is there an assumption that most users will have a mix of data which is readily compressible and hence the manufacturers can quote compression ratios?

    1. Arglebarglewargle

      Re: Compression question

      Many moons ago, I worked as a backup sysadmin.

      The backup system we were using would estimate tape capacity based upon "compressed" capacity - so for an LTO4 cartridge, for example, it would say "1.6 GB". It would display the percentage full based upon the amount of data written, divided by that estimated capacity.

      If the data was REALLY compressible, it'd hit the 1.6 GB mark, say it was 100% full (in a "filling" state), and keep going. It wouldn't mark the tape "full" until it actually hit the end of the tape.

      Conversely, if the data was not compressible at all, it'd hit the 800 GB mark, end of tape, oh, the tape's full, we've just jumped from 50% of capacity to 100% in a matter of seconds. My bad, blush, so sorry, I'll try not to do it again.

      In other words: the software you're using, if it's properly written, will handle it all without any issue. It just means that you hit the end-of-tape marker (meaning that the tape is full and can't take on any more data) sooner than you might expect; any sane software that uses tape knows how to handle end-of-tape without breaking, regardless of when it happens.

      Compression ratios are basically just the manufacturer's way of inflating their specifications without having to be held to account if said specifications aren't met. I ALWAYS looked at tape's native capacity, ignoring the drive's compression algorithm, in sizing things - the compression was a bonus, and data growth over time would soak up the excess anyway in fairly short order. (Seriously: in two years at one job, a core database went from about 400 GB to about 4 TB. A ten-fold increase, in just two years, and it was STILL GROWING. I was extremely grateful when I moved on and didn't have to deal with that particular issue any more...)

    2. hellwig

      Re: Compression question

      I have the same thoughts.

      If you use tape backup for, say, text databases, it might be very compressible (3:1 might be an underestimate). If you use it for binary media (videos or something), you probably won't get any compression.

      The compressed size is really meaningless, and I wonder if anyone actually uses that.

      Imagine if HDDs or SDDs were sold in that same fashion. Or maybe clothes: "Sure, this is a 32-inch waste, it could accommodate and 36-inch stomach, if you suck-in your gut."

  9. Anonymous South African Coward Bronze badge

    Fond memories of listening to the whirring sounds of a 250Mb tape drive running from the floppy controller. Back and forward whirring all the while :)

  10. JeffyPoooh
    Pint

    A box of USB Hard Disk Drives

    That's all that's required for most. The rest is just some policy decisions and a few lines of code.

    Acknowledge in advance that if you're actually generating 15TB of new data every 24 hours, then perhaps you'd need tape backup. Normal office servers certainly do not.

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