back to article No TKO for LTO: Tape format spawns another 2 generations, sports 120TB bigness

The LTO consortium has added two more generations to its tape format roadmap, extending it out to LTO-10 and a 120TB compressed capacity. The roadmap used to finish at LTO-8, and that version of the LTO tape format future was announced in April 2010. Now, four-and-a-half years later, it has been extended again. LTO …

  1. Buzzword

    Back to maths class!

    "Basically raw capacity is doubling every generation with compressed capacity increasing 2.5 times per generation."

    No, that's not mathematically possible. Compressed capacity increases at the same rate as uncompressed, assuming no change in compression efficiency. And indeed compression has pretty much stopped evolving: all the low-hanging fruit has been plucked, there are only minimal gains left to be made.

    What's worse, a lot of the new data which is filling up these tape drives is made up of images and video, already compressed: so the stated 2.5:1 compression ratio is actually falling.

    1. phuzz Silver badge
      Thumb Up

      Re: Back to maths class!

      I'm not sure why you got a downvote there as you're entirely correct.

      For reference, at my old job we used to get compression rates of around 1.5-2:1 on user files (Word and Excel docs, pst files etc.).

      1. Riku

        Re: Back to maths class!

        Curious - was that before or after MS added in-file compression? (Office 2007-ish?) We used to get great compression on backups but now alomst everything has some sort of self-compression built in to the format. If you're still getting 2:1 on the new self-compressed files, that's pretty good.

        1. phuzz Silver badge

          Re: Back to maths class!

          It was around the time we introduced 2007, so most of the documents would have been older versions. I don't work at that job any more so I'm not sure what they're getting now.

  2. Warm Braw

    Lowest-cost archive medium

    While tape might still be cheapest, the margin by which it is cheapest keeps dwindling. There was a time when the cost of backup media was around 10% of primary storage of a similar size, it now seems to be more like 30%. If it creeps up much higher, the inconvenience factor becomes a significant consideration.

    Not to mention that if you really want to "archive" something for a prolonged period of time and have it readable by future generations, you're still better off sticking to ink on paper.

    I'd like to see cheap and durable archival storage on the list of candidate projects for the longitude prize - we're going to need it.

    1. Charles 9

      Re: Lowest-cost archive medium

      I would love to have even something of an LTO-6 level of capacity, but at consumer prices (which they're decidely NOT--an LTO-6 drive runs nearly $6,000). Having something that can several TB of things pretty safe for the mid-term, say around five to seven years, would be really nice for packrats such as myself. Right now, external hard drives remain the most affordable choice in the consumer market, but I still have to wonder about their reliability and data retention at these lengths of time.

      "Not to mention that if you really want to "archive" something for a prolonged period of time and have it readable by future generations, you're still better off sticking to ink on paper."

      Kind of hard to put a movie on pen and paper. Same for a selection of music (sheet music is basically musical source code--most people prefer finished products).

      1. Joerg

        Re: Lowest-cost archive medium

        LTO drives are not consumer products. The drives are for the high-end enterprise market.

        Anyway on www.pricewatch.com LTO-6 drives (not tape libraries) are in the $2,000 range right now..

        Like:

        Hewlett Packard - HP LTO-6 SAS 6250 Internal TAPE DRIVE S-BUY EH969SB

        $2047.00 $2047.00 w/ship Ship: FREE ONLINE

        Quantum - LTO-6 HH Tape Drive, Internal Kit, 6Gb/s S LTO-6 HH Tape Drive Internal Kit 6Gb/s S

        $2187.21 $2194.49 w/ship Ship: $7.28

        1. Charles 9

          Re: Lowest-cost archive medium

          I know isn't a consumer-level product. I'm just pointing out there is also a need for backup media on the consumer end, too (indeed, many would say it's underserved). Consumer drives have reached multi-terabyte levels, and people are filling them up. Optical discs are rapidly being left behind in that regard; not even the upcoming Archival Disc will back up a 1TB HD in one disc, and the next step down, BD-R, is way too small. And there hasn't been a single consumer tape improvement since Travan-40 (raw capacity smaller than a BD-R). Which means pretty much the only practical way to back up a hard drive full of data is with another hard drive. But the reliability of external hard drives can be inconsistent, raising the specter of a Failsafe Failure before a cycle change occurs (I just did one when I transferred out my hundreds of backup DV-R's to HDs, and that wasn't without sporadic losses of data). So pardon if I seem a little concerned about mid-term data retention on the consumer front.

          1. NinjasFTW

            Re: Lowest-cost archive medium

            This makes me wonder if there might a business model for a consumer level tape backup service.

            You register, I send you a prepaid mail back which you put your hdd/dvd/bluray etc and send in.

            I move all the data to tape and send you back the tape with an index file of everything that has been copied over.

            Restores work in the same fashion.

            Yes its a slow service but I would take a yearly option to have that done for piece of mind.

            1. phuzz Silver badge

              Re: Lowest-cost archive medium

              I guess that's almost where Amazon Glacier is aiming for.

              Currently it's supposed to be business only, but I'm sure you could get them to archive your data to tape by throwing money and a bit of pretending.

              (edit, looks like I'm not the first to have that idea:

              http://lifehacker.com/how-to-use-amazon-glacier-as-a-dirt-cheap-backup-solut-1460814873)

              1. phil dude
                Boffin

                Re: Lowest-cost archive medium

                economically there might be a case - if I understand correctly, the manufacturers guarantee one generation behind can read/write and two can read?

                Hence, if a service put stuff on a tape for you , you could get a (I assume) much cheaper drive from 2 generations behind to read it...

                P.

                1. Fat-Boy-R-Dee

                  Re: Lowest-cost archive medium

                  Um, not quite.

                  Drives cannot read uplevel media (for example, an LTO-5 drive cannot read LTO-6 media).

                  Drives can read and write one downlevel of media, and read two downlevels of media. For example, the same LTO-5 drive can read LTO-5/4/3 media, but can only read/write LTO-5/4 media.

          2. Riku

            Re: Lowest-cost archive medium

            There is, it's called DAT. problem is, most *consumers* say things like "USB sticks are cheaper than that. yes, they are, and good luck to you getting your data back in five years from one that was a freebie at a trade-show. "Prosumers", maybe they will buy something like a DAT drive. probelm is today, DAT is no longer being scaled by HP and LTO's underlying technology is just too expensive to send the older generations downmarket. (Those LTO tape heads cost a fortune to manufacture).

            Consumers want the same features (in this case the reliability, durability and low likelihood of bad bits), but they don't_want_to_pay_for_it. End of story. Having worked in an R&D facility, despite what everyone thinks, this stuff does cause the scientists a lot of headaches and sleepless nights and does cost a lot of real, hard cash that has to be recouped. Consumers just don't want to admit that - a behaviour somewhat akin to El Reg's "freetards".

            Enterprise-class features *may* eventually get cheaper, but in order to do that, they have to acheive scale and some healthy competiton helps too. With some very cool technolgies, thus doesn't always occur.

            1. Charles 9

              Re: Lowest-cost archive medium

              Anyone interested in the consumer market has to get used to the idea that price matters. It's not so much "We don't want to pay for it" as "We can't pay for it." For the consumer, "You Get What You Pay For" only goes so far, especially with limited budgets and competing interests. That's why there's the concept of the "comfort zone" beyond which any attempt to woo the customer will fail to attract all but hardcore adherents. The demand curve for the consumer is necessarily low and shallow. If the tech is such that even the lowest end is too expensive, that means supply and demand can't meet, leaving an untapped market.

            2. Trygve Henriksen

              Re: Lowest-cost archive medium

              Please, NO!

              DO NOT mention DAT and backups together!

              The only media worse than DAT is VHS tapes. Even 8mm tapes were better!

              DAT tapes can't F! handle the all the sudden starts, stops and winding operations and will stretch the band, making it unreliable after just a few uses.

              I've used them all, DAT, Mini DV, 8mm, QIC, SLR, DLT, super DLT, LTO2, LTO3, LTO4...

              And probably a few I've managed to repress the memory of...

              For home use today, I'd say to use multiple USB or networked HDDs, do 'full backups' and not to trust a drive older than 3 years.

              I have a 'main' 4TB networked drive at home(locked in a closet together with the router), and two USB drives of the same capacity. Every month I backup the networked drive to one of the USB drives, and bring it 'offsite(drawer in my office), then I bring home the other drive to await the next 'full backup'

              Any changes between the backups is copied to a USB-stick.

              In the case of many changes or lots of files added(ripping a complete Dr. Who season from DVD, maybe) I do the backup as soon as the changes are done instead of waiting for the 'monthly'

              1. Down not across

                Re: Lowest-cost archive medium

                Please, NO!

                DO NOT mention DAT and backups together!

                The only media worse than DAT is VHS tapes. Even 8mm tapes were better!

                Hear hear! I've had lots of different DAT drives and they've all been unreliable. Pretty useless really.

                Whereas Exabyte (8mm), DLT/LTO have performed very reliably. Even QIC (I'm talking about DC600 and DC6150 cartridges here...) was fairly realiable. I did briefly try Tandberg's SLR and it seemed OK too, but didn't use it long enough to make proper judgement.

          3. David (procurepod)

            Re: Lowest-cost archive medium

            it the lowest yet more compatible medium For storage data uptill now, what i have seen is that this is the most reliable medium in offline data storage too.

  3. Anonymous Coward
    Anonymous Coward

    2.5:1 compression is just misleading

    The only relevant metric is raw capacity. Today, when you encrypt and compress data before backup announcing tape with official capacity computed with 2.5x compression ratio is misleading. I'd understand 2:1 for historical reasons (every vendor did that, so user just divides it by two to get baseline for his own computation and all is well). 2.5 ratio is just trickery, it does not make any sense.

    1. Joerg

      Re: 2.5:1 compression is just misleading

      Yeah.. there is marketing.. BUT .. it's not that uncompressed 26TeraByte at 700MByte/s speed for LTO-9 tapes and 48TeraByte at 1.1GBytes/s speed for LTO-10 are going to be disappointing at all... !

      That is more than enough capacity and performance for the next 10 years.

  4. picturethis
    Childcatcher

    Right, good up until the fact that only LTO-6 exists..

    I was excited to see this up until the point where the article states that everything but LTO-6 is vaporware/wishware (the hardware equivalent for vaporware).

    I've learned over the (20+) years not to make decisions or depend on anything that I can't currently buy. The future is too unpredictable where tech is concerned.

    By the time the newer formats are available, the average disk size will probably be 1000 TBs or greater. So what this article needs to do is create a chart that maps planned future tape storage capacity along with planned future hard drive capacity to see if things are getting better in terms of being able to use tape. Is tape winning or losing the war? I suspect it's losing, as it has been for many years. This was the entire reason tape libraries needed to be invented.

    Sorry, but my backup needs are now, not something that HP or IBM or whoever <may> offer years from now to solve.

    Backups of Backups using hard drives is about it for now, even for us higher tech consumers. I have many hard-drives (some of them ATA33/IDE) that are still readable even after 10+ years and that's without taking any special environmental care of them. Although I will admit that I haven't checked every file on them..

    Of course for the lazy, there's always the Cloud.... Maybe someday Cloud providers will provide backups for each other as the final solution. Ultimate redundancy :)

    1. phuzz Silver badge

      Re: Right, good up until the fact that only LTO-6 exists..

      Well there's the basis for an article; What have the LTO group promised in the past, and have they delivered?

      As far as I know they've always managed to hit the capacities and speeds that they've planned for LTO-2 to 6, but maybe not the release dates.

    2. Joerg

      Re: Right, good up until the fact that only LTO-6 exists..

      What? Are you serious?

      LTO-7 is going to be released really soon now. Products should hit the market in a few months.

  5. Anonymous Coward
    Anonymous Coward

    LTO-9

    We are all interested in the future, for that is where you and I are going to spend the rest of our lives.

  6. Anonymous Coward
    Anonymous Coward

    Amount of time to back up to tape

    All this extra capacity is great, but how long does it take to write 2.5TB (LTO-6) of data to tape?

    What data interface would you use to keep the drive constantly streaming to prevent start/stop actions?

    1. Joerg

      Re: Amount of time to back up to tape

      LTO-6 drives are either SAS or FC (some expensive models, usually tape libraries, use Ethernet with link aggregation support).

      For LTO-6 the uncompressed native transfer rate is sustained 160MB/s when reading as well writing.

      Which means a 2,500,000MByte tape (actually just like with HDUs and SSD and CD/DVD/BD the real size it's base 2 not base 10...but then marketing... ) as written on IBM website: http://publib.boulder.ibm.com/infocenter/ts3500tl/v1r0/index.jsp?topic=%2Fcom.ibm.storage.ts3500.doc%2Fipg_3584_meltoworm.html

      "Ultrium 6 2500 GB (2328.31 GiB)"

      So it means 2,328,310 MB / 160 MB = 14551.9375 secs = 242 minutes = 4 hours 2 minutes approx

      We can expect LTO-7 6.4TB tapes then to take: 6,250,000 MB / 315 MB = 19841 secs = 330 minutes = 5 hours 30minutes approx

      For up to LTO-4 120MB/s standard consumer grade cheap SATA 7200rpm Hard Drives currently sold are more than enough .. even LTO-5 at 140MB/s could work...

      For LTO-6 at least use two consumer grade drives in RAID-0 better on an hardware RAID controller anyway.

      Obviously for professional use it's common to use no less than 8 HDUs in RAID-5 or RAID-6 .. otherwise for 16+ drives RAID-60 it's a better option. With hardware controllers supporting SSD caching then adding up to 4 SSD MLC or SLC drives for speeding the arrays up even further it's a good thing.

  7. DainB Bronze badge

    It's all about size

    Try move 100TB of data between continents in couple of days and you'll soon realize that cheapest, fastest and the only reliable way to do it is put it on 2 tapes and ship it with courier. Now try do the same with how many, 20-40 hard drives ?

  8. itzman
    FAIL

    success rates in restoring from tape?

    back in the day, about 25%.

    3/4 of all tapes were unreadable after a few years.

    Or even a few months.

    I'd rather have my rust spinning or still in a sealed enclosure, with the same read heads as what wrote it. than in a flipping tape cartridge.

    So its continuously replaced mirrored disks for me.

    As for moving 100TB across an ocean, well the Internet was supposed to mean you didn't actually have to.

    And in any case 100TB in 48 hours is only 4.6Gbps.

    Well within a single fibre capability.

    And that way it gets loaded direct onto the disks at the far end without a tape reader.

    That will probably fail to read it anyway.

    1. Trygve Henriksen

      Re: success rates in restoring from tape?

      Which tape formats did you try to recover from?

      It sounds like DAT, MiniDV, or possibly a 40GB DLT...

    2. DainB Bronze badge

      Re: success rates in restoring from tape?

      "Well within a single fibre capability."

      If you have 10Gb dark fiber from Oz to USA for example and chances of that are miniscule.

      4.6Gb/s over Internet ? You must have some very special internet.

    3. Joerg

      Re: success rates in restoring from tape?

      You were using old digital tapes technologies, clearly.

      LTO tape drives have a UBER 10E17 reliability. There is a lot of redundancy on the tapes, in the 30-40% range at least.

      The tape must be seriously damaged for the drive being unable to restore data from it.

  9. Anonymous Coward
    Anonymous Coward

    Capacity increasing, speed not so much?

    I wonder what will happen when you need to backup, say, 2 YB of uncompressible data, and it will forcibly take 1 week to accomplish said backup, and new 2 YB of uncompressible data is produced every week. Will you stripe together LTO tape writers as it was a RAID setup?

    I'm exaggerating on the prefix on purpose (yottabytes are a hell lotta bytes), to make people think of what will happen when backups can't keep up with the data generation business that you want to backup.

    By the way, googling the term showed the NSA is already aiming at that volume of "Yottabytes"... 10^24 bytes for a datacenter.

    1. Joerg

      Re: Capacity increasing, speed not so much?

      Where money is not a problem for large multinationals and security agencies... they have no problem buying very expensive, big, really massive LTO tape libraries with RAID functions included too.

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