Slashdot Mirror

← Back to Stories (view on slashdot.org)

Ask Slashdot: Personal Tape Drive NAS?

Posted by timothy on from the can-tape-actually-be-that-fast? dept.

New submitter hey_popey writes "I would like to piggyback on a previous Ask Slashdot question. Do you know of any realistic way to use a tape drive solution at home, not as a backup, but as a regular NAS? I would like, for example, to save the torrents of my Linux distributions on it, and at the same time, play the family videos on a computer. It would seem at a first glance that the transfer rates and capacity of Linear Tape-Open (1.5TB, 280MB/s in 2010) and the functionality of LTFS would allow me to do that, but I don't know the details, or whether this would be economically viable."

10 of 268 comments (clear)

  1. Nope. by Sprite_tm · · Score: 5, Informative

    The big disadvantage of tapes is that it has long seek times. Not 'long' as in a few times that of a hard disk, but 'long' as in: can take a full minute to do. Access of multiple files on a normal HD is done by reading a meg of the first file, then seeking to the second file and reading a meg, going back to the first file and reading a meg etc. On a tape drive, even when the seek time is only, say, 10 seconds, you'd get a total throughput of 100K/sec that way. And I'm not even talking about the havoc that using it for storage of torrent files wreaks on it: that's a random-access process if I ever saw one, and the seek times on tape would kill your bandwidth very quickly, and probably your tapes too (because of wear&tear).

    1. Re:Nope. by Lumpy · · Score: 5, Informative

      Yup a lot of people don't remember linear access. and a minute or two? I remember a 15 minute seek and load time when restoring a file from the end of an archive tape.

      In fact it's faster to download a new distro on a 7mbps DSL line than it is to find it on the tape.

      Now, I have seen with a tape robot cabinet a "infinite" hard drive. 4 hard drives for online storage, they had 4 hard drives for nearline storage, and all the tapes in the cabinet for offline storage. if you accessed the file from online and it was on tape, you would get a winpopup from the server stating that the file is in offline storage and will be spooled up for you. it then would email the person when the file was put back into nearline or online storage storage.

      back in 2002 it was how we had 22Tb of tv commercials, Tv show productions, and video footage available for the video editing suites.

      --
      Do not look at laser with remaining good eye.
    2. Re:Nope. by fuzzyfuzzyfungus · · Score: 4, Informative

      and probably your tapes too (because of wear&tear)

      the hideous, desperate seeking of a tape in this condition is informally described as 'shoe-shining'...

    3. Re:Nope. by FrederikNS · · Score: 2, Informative

      Hypothetically a 700 MB linux iso would take 13 minutes and 58 seconds to download on a 7 mbit/s connection. If your tape is in position; then of course it would be faster. But in case that you have to seek to the position of the iso, and you have a 15 minute seek time as Lumpy mentioned, plus the 3 seconds it would take to read, then downloading the iso is still faster.

    4. Re:Nope. by tibit · · Score: 3, Informative

      All I remember is that my tape adventure ended with an LTO-3 drive that lasted about a year, doing daily backups, on a 7 tape daily rotation. I had two LTO/DLT generations before that, and a prior helical scan system with smaller tapes as well. The Linux kernel drivers for tape devices never worked all that well -- getting good throughput required tweaking and I had to add a large buffer between gzip and tape in spite of a fast machine. The particular drive in question worked fine on RHEL4 but would not work on a Dell server under RHEL5 where the throughput went to hell -- it was 5-8x slower. If you plan on using any tape drives, make sure that you test the exact combination of server, interface, interconnect, drive, tape, Linux distribution and backup software/scripts you're going to run it as. Anything less may lull you into a false sense of accomplishment that will be blown away in short order once it fails in production.

      In times of linux kernel 2.0 I had some junk machines that had ISA-based Adaptec SCSI cards and worked quite well with old 80 megabyte aluminum-plate-backed tape cartridges. It was slow, but it still kept the tape streaming at full speed. It was foolproof, and redirecting tar -z output to /dev/st0 was all that was needed to get it to work, after setting desired block size. Even on a 486-class machine. It seems that it was downhill ever since in terms of ability for the systems to maintain streaming and retain data. I can still read those 80 megabyte tapes with zero errors. All of a dozen of them or so. All of the newer tapes would develop errors under normal use, sometimes after a couple of uses!

      --
      A successful API design takes a mixture of software design and pedagogy.
  2. Cost vs HDD Solution by thesandbender · · Score: 5, Informative

    The overwhelming issues with latency aside, a 1.5TB (native not compressed) LTO drive will set you back ~1800 USD and you'll need an extra ~100-150 for a SAS controller that can drive it. For that price you can by yourself 24TB of HDD storage (12 x 2TB) with enough money left over for a decent SATA/SAS RAID controller. If you setup a RAID 10 array you'll have 12TB exponentially faster access times and better data security (unless you make copies of every tape).

    1. Re:Cost vs HDD Solution by dtdmrr · · Score: 4, Informative

      Note that the 1800 is just for the tape drive. An 8 tape library with drive and media will be more like $4k, and that still only gets you 12TB (given the file types you mentioned, don't plan on getting any capacity boost from the LTO compression). You will have to go with one really big library before tapes win on price. Unless of course you are willing to change tapes manually, or build your own robot/library out of lego. But even then that 24TB figure is only a lower bound on the cross over.

    2. Re:Cost vs HDD Solution by thesandbender · · Score: 5, Informative

      It's economies of scale, tape has a high cost of entry but a relatively low maintenance cost. A 1.5TB LTO 5 tape costs 40 USD. A 1.5TB drive costs 90 USD. The VM enclave I use for testing at one client has 700TB, to back up that data set with HDD would cost 23,333 USD more than tape (for just the media). That difference alone covers the cost of a tape library. And, most corporations are going to take complete backups once a week with incremental backups during the week. Which means an extra 23,333 a week (HDD vs tape). Scale this out to petabytes of data and HDD's become prohibitively expensive.

      Also, one of the primary reasons to use tape is you can store them offsite for disaster recovery. You can put a box full of tapes in the back of a panel van and drive them down a bumpy gravel road without any big worries, you just can't do that with HDD's with out protective housing.

  3. Tapes. Are. Useless. by fraxinus-tree · · Score: 5, Informative

    They are not only useless at home. They are completely useless as a backup solution in the first place. They refuse to read in 95% of their intended usage scenarios, including, but not limited to, incompatible/failed tape drives, missing/obsolete/buggy/outright stupid software, degraded/stretched/torn off tape, mislabeled/misordered media and so on. And then again, they cost $$$$$, because PHB's keep on buying them. And they do, because they like solid-looking stacks of backups. Even if no one prescribing them in the backup plans had ever tried to restore a single file in the last 20 years. Or ever.

    Hard disks are good. They are also good for backups. They are cheap, they sell them in the shop down the street, they work 99.99% of their intended usage scenarios, do very well in every other usage scenario, and they can be easily connected to any computer, just to see what's in.

    1. Re:Tapes. Are. Useless. by Anonymous Coward · · Score: 5, Informative

      They refuse to read in 95% of their intended usage scenarios, including, but not limited to, incompatible/failed tape drives, missing/obsolete/buggy/outright stupid software, degraded/stretched/torn off tape, mislabeled/misordered media and so on. And then again, they cost $$$$$, because PHB's keep on buying them. And they do, because they like solid-looking stacks of backups. Even if no one prescribing them in the backup plans had ever tried to restore a single file in the last 20 years. Or ever.

      Being an enterprise backup engineer, I have to disagree with the bulk of this. We do thousands of tape restores every year, the tape media is extremely reliable, and modern tape is FAAAST. Nothing can do streaming database backups or restores faster than tape, particularly multithreaded - at that point, most servers and network infrastructures don't have enough throughput to keep up with the tape drives.

      Now, Joe Admin at an SMB with hand-labelled tapes (versus barcoded), poorly managed and mismatched tape drives, and poorly planned and tested emergency recovery scenarios is a problem.... but that just means that the person making the technical and purchasing decisions doesn't know how to take advantage of tape properly. If you've got mismatched tape hardware versions, tape drive firmware versions, software problems, etc - well, that's a problem of the person managing the solution. Like any other technical solution, if you're poorly trained on backup solutions, you're going to have a poor implementation. You reap what you sow.

      Disk backup has its place, certainly (and my company leverages it heavily as well), but don't write off tape as useless or a poor technical solution. If you need a database backup restored with the smallest RTO, nothing does it faster than tape.