Magnetic Tape Storage May Not Be Retro

Magnetic storage is quickly becoming an antiquated technology but IBM may have given it a few more years. Currently, magnetic storage is still manufactured as hard disk drives (HDDs) but you won’t find a tape drive in a modern consumer computer. That’s not likely to change but IBM is pushing the envelope to make a tape drive that will be smaller and more economical than other massive storage options. In many ways, they’re the antithesis of solid state drives (SSDs) because tape drives are slow to retrieve data but capable of holding a lot inexpensively.

Three advances are responsible for this surge in capacity. Firstly, the tape “grains,” where each bit is recorded, have been shrunk by sputtering metal to a film instead of painting it on. Secondly, better servo control allows the reading mechanisms to read those tiny grains with the necessary accuracy. Lastly, stronger computation is used to read the data by using error detection and correction because when your tape is traveling four meters per second, it takes a long time to go back and double-check something.

IBM’s tape drive won’t replace your hard drive but it could back it up daily, many times over.

Check this out if your wetware needs a memory boost or this if your breakfast needs a memory boost.

64 thoughts on “Magnetic Tape Storage May Not Be Retro

  1. I’ll keep an eye out for that, in about 2040 when I build my “Ultimate 2017 vintage system” with an array made out of those 60TB SSDs I picked up at yard sales. Then at least I’ll be able to back it up.

    1. In 2040 quantum teleportation will make any “attachment” of your machine to a storage device which you host obsolete with instantaneous transmission of data to and fro. What this means is that your “device” … which by the way is likely to be embedded into your brain by this time … will have storage and processing and all of that somewhere else. You will think you are having fun … it will feel real. However, you will be working in the mines. We will be in the virtual world … having a hamburger, fries, and a vanilla malt … perhaps it’s nested already … or perhaps the tech already exists. You can take this to many intriguing and thought provoking levels.

  2. Large enterprise IT still happily uses tape for long term storage, because given the volume of stored data, the expensive tape drive mechanic will pay for itself rather quickly and since it’s backups, the seek time is not that much of an issue…

    Last but not least, the tapes themselves are surprisingly resilient and currently are a probably the best choice for high capacity, (very) long term storage.

    1. *looks over at stack of LTO4 tapes*

      Yeah, tape is “retro.”

      Stop trying to be a shitty technology blog, HaD. Just post people hacking stuff. I know from friends that you get tons of submissions, a ton of which you ignore. And yet you post this shit.

    2. Yes. I’ll never forget when EMC pitched us a disk to disk backup solution….. We kept preaching we’d need the cost/scalability (in terms of keeping our required x amount of dailies, monthlies and yearly backups). In the end they brought us a quote back that was like 3x our current SAN cost… Politely showed them the door and installed an updated robotic library and all new tapes at the current LTO gen that was 1/100th the cost….. Sigh….

    1. I recently recovered data from a tape (1600BPI) written in 1985. Storage of the reel over the years was, to say the least, suboptimal. The hard part was finding a drive that could read it. Read clean 30+ years later.

      1. “The only problem I ever had with tape backups were consumer grade floppy interface tape drives, like cheap QIC-40 drives. And even then it was *usually* user error.”
        I would suggest the most likely cause of problems with tape systems was dust. Most tape systems relied on optical sensors to detect things like tape start/end, and dust and dirt would render them inoperable. Dirty tape heads came second on the list.

  3. I’d be very interested in the durability of vacuum coated polymer magnetic media. This process is done all the time with plastic films using aluminum (that cheap emergency blanket or shiny bag of chips) but it’s not very durable – it can rub off easily and will deteriorate chemically almost immediately under the right circumstances.

  4. “…but you won’t find a tape drive in a modern consumer computer.”

    Historically I would argue one rarely found a tape drive in a “consumer” computer. I think you need to do a bit more research. I use LTO-4 to 6 all the time for small business backups and my own personal backups in my home machines – depending on needs and what is price appropriate. It is a direct technology evolution of what met the use cases floppy-tape, QIC, DAT, and other formats previous filled in the 80s and 90s. LTO is an apples to apples comparison of those old formats. Tape hasn’t died in 2017. LTO-7 is about to be released and LTO-8, 9, and 10 planned.

    https://en.wikipedia.org/wiki/Linear_Tape-Open

    1. Those of us with experience in IT or server management are well aware of tape’s continued use for backups, but what I believe Brian was referring to with that line is the kind of cassette drives used in the very early desktop machines.

    2. The Coleco Adam was definitely a consumer computer, and it had an actual integral tape drive, under control of the computer, as opposed to interfacing with an external, human-controlled audio cassette player. Much faster than audio tape but still miserable compared to a floppy drive.

  5. At a storage conference recently I heard the IBM tape guys say something to this effect: HDD technology is hitting a wall which is defined by physics. Flash will be hitting similar physics walls within a few years. Tape’s limitations are all engineering problems and won’t hit any physics barriers for a couple of decades.

    1. Current flash technologies have already hit some of those walls, cell size vs. durability being the most obvious. Hence the push for ‘multi-layer’ devices in an effort to raise per device capacity.

    2. Sounds like bull to me. As both HDDs and tape use magnetic storage they are experiencing similar physical problems for the actual storing/retrieving of data (size of magnetic domains, leakage between adjacent data fields etc.) with similar ways to handle them (change of magnetic media, addition of pattering, going perpendicular*).
      There are of course different problems that aren’t shared. Tape can stretch while the rigid platters of HDDs don’t (but they can of course expand when heated). HDDs need a gas for the heads to float on while keeping a ever decreasing distance from the platters.

      But perhaps the lack of physical limitations for tape was the fact that one could theoretically expand the size of a tape to be solar system sized or more.

      (* https://www.youtube.com/watch?v=xb_PyKuI7II )

      1. The limitations they were referring to were specifically bit cell size (i.e. areal density).

        I found a set of slides from 2016: http://www.digitalpreservation.gov/meetings/DSA2016/Day1/Decad_TAPECLOUDArchiveStorage__LOC__Decad_final.pdf

        Assuming their numbers are right, I think they may be on to something: A NAND bit is around 17.3nmx17.3mm. A HDD bit is 47nmx13nm and an LTO7 tape bit is 2850nmx52nm. There is a ton of room to grow (ie shrink the bit size) before the same sort of problems hit tape.

        The bit error rate slide is interesting too

  6. What’s needed is an affordable LTO SATA drive for desktop computers. There are ones made that will fit into a normal 5.25″ half height drive bay. But if you want the latest version it’ll cost as much as a decent used car.

    In contrast, the tapes are fairly inexpensive. If IBM wants LTO to have a larger market, they need to get a consumer version of a drive out the door. Such drives won’t need to be able to run 24/7, they’ll see far less uses than drives used to backup datacenters or even single servers.

    What would be better than LTO? A WORM version of Fluorescent Multilayer Disc. That technology passed through a few hands, eventually seeing production of a couple of player models and a small selection of movies in Asia. Those had a capacity somewhere between HD-DVD and Blu-Ray but the tech has the capability of being expanded to much more than the latest revision to Blu-Ray. The ‘problem’ with FMD is none of its owners belonged to the DVD Forum or DVD+RW Alliance groups.

    1. Whenever I’ve looked at tape drives I’ve not figured out been puzzled as to why they are all SCSI and not SATA – seems like one of those “nobody makes it because nobody asks for it”… “nobody asks because nobody makes it” situations.

          1. @jacques – depends on what your paying for tape. My quick search shows lto 6 media is significantly cheaper. You’d be paying close to a grand for the same capacity you can get for $30.

    2. DVDRW ate tapes lunch for a while, but even before BDROM was out, consumer HDD sizes had blown way past it’s nominal capacity. You need more than 2 pieces of media for backup, you tend to say screw it. So I don’t really understand why there hasn’t been a good, large capacity consumer backup system.

  7. Tape is still used today not only for backups, but also for archiving and restoring large quantities of data. Working at a TV station, we archive all sport at 120Mbps, which means a 2 hour event weights more than 100GB. As we record 30 or 40 events per week, we would need to add a 4 TB hard drive to the system every week (and that system is running for 6 years now).
    WIth our tape library ( https://www.oracle.com/storage/tape-storage/sl8500-modular-library-system/index.html ), archiving from and restoring to our NAS is a seamless operation.
    As for “I never trusted tapes”, errr, I guess you have not used the right tape system :-).

    1. Ah 120Mb/s… Probably DNxHD in 8 bit? … Which means I can probably guess the video server and the MAM.
      We also go LTO for our sports archives. Somewhere in the order of 15TB / week. (don’t quote me on those numbers, they are rough calc after lots of beer).

  8. Tape has never been retro when you need extreme data volumes (backup (google use this and I’m certain most others), rarely accessed vast quantities of data (scientific data from particle accelerators and similar))… At the end of the day for volumes like this it comes down to price per TB (and reliability), where you simply can’t beat tapes (to the best of my knowledge). As a bonus – no power is needed for petabytes of storage, where as the HDD/SSD equivalent would be pretty power hungry. And finally – tape libraries look awesome!

  9. Sony is trying to push its Optical Disc Archive solution as a competitor to LTO:
    https://www.pro.sony.eu/pro/lang/en/eu/products/archiving-storage-optical-disc-archive
    3.3TB per cartridge is current max capacity, but for a “Write Once” only cartridge.
    There are several discs inside a cartridge, and the current drive has 4 heads/8 laser channels. Read speed is 250MB/s and write speed is 125MB/s (for “write once” cartridge, slower for read/write ones).
    Latest USB 3.0 standalone drive is $7500 US list price, and 3.3TB write once cartridge is around $150.
    There’s also a library version (base + expansion modules).
    https://pro.sony.com/bbsc/ssr/cat-datastorage/cat-opticaldiscarchive/
    Impressive piece of technology, but from my point of view: too expensive, too slow and too proprietary (Sony is the only company to build them and can stop them at any time, like they did with the DTF archive tape format based on Digital Betacam 1/2″ inch tape format).
    If it were really low cost, could be more interesting for short/mid-term storage.
    Also, cartridge life is announced to be 100 years. Good. But who will have a working drive, and compatible and working PC and software in 100 years?
    What do you think?

  10. You know what would be cool? Is if they could make a 2 Terabyte “stringy floppy” with this. I think the numbers work. Question is, time to market, if they take so long to intro it, that ppl be getting 10TB drives as default, then less interesting.

  11. I’ve seen a demo of the EverSpan version of this product. When they mentioned the 100-year lifespan, they also added that there would be no reason to verify the media. I called them out on it — people were told the same sort of thing when CD-R and CD-RW were the rage and since one cannot prove a negative, it is imperative to continue to check media fixity periodically.

    Single vendor is a real issue as well. IBM has pretty much cornered the market for tape drives (LTO and enterprise), but with IBM, as long as you’ve got the cash, they’ll continue to support it.

  12. “Scientists at IBM Research say they can now store 201 gigabits per square inch on a special “sputtered” tape made by Sony Storage Media Solutions.”

    Really? No numbers in the summary at all?

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.