Stories
Slash Boxes
Comments

SoylentNews is people

posted by n1 on Monday May 05 2014, @12:01PM   Printer-friendly
from the sometimes-a-good-idea-is-a-bad-one dept.

With the upcoming release of Mint 17 LTS, I am looking to upgrade from Mint 13 primarily because of the new BTRFS file system. I have an older netbook with very limited hard disk space that just won't die, so a compressed file system is very attractive to me.

I've found instructions how to do it so the only issue is whether to go ahead... This is where I ask for your wisdom!

This discussion has been archived. No new comments can be posted.
Display Options Threshold/Breakthrough Mark All as Read Mark All as Unread
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
  • (Score: 5, Insightful) by The Mighty Buzzard on Monday May 05 2014, @12:20PM

    It's not especially technically difficult but I wouldn't. The main problem with laptop performance, unless you're using an SSD, has always been and is always going to be I/O to disk storage. Compression adds even more latency. For me that would be the straw that tips barely usable to unusable.

    It also adds cpu overhead, which will drain your battery quicker, so that's a consideration too unless you're always running off AC.

    --
    My rights don't end where your fear begins.
    • (Score: 3, Insightful) by damnbunni on Monday May 05 2014, @12:28PM

      by damnbunni (704) on Monday May 05 2014, @12:28PM (#39744) Journal

      I've seen that work both ways though. If the hard drive is especially slow, compression can actually speed things up; the CPU spends time to compress the data, but the smaller amount of data being written takes less time.

      It really depends on just how slow the disk is.

      And sometimes even if compression DOES slow things down it's still worth it. Like on my Asus Eee 901. It has a 4 gig C: drive. So yes, I turned NTFS compression on for most of the files on it.

      • (Score: 3, Informative) by Bill, Shooter Of Bul on Monday May 05 2014, @02:22PM

        by Bill, Shooter Of Bul (3170) on Monday May 05 2014, @02:22PM (#39796)

        Yeah, it depends on the CPU speed vs Disk Speed. In most modern computers with overly fast cores and a number of them sitting idle, the minute delay in compressing the data is made up by the time saved by writing smaller chunks of data.

        I know phornoix articles aren't always super scientific, but the test suite is a good idea for testing the same system with two different layouts. I'd say try it both ways, benchmark it for us and let us know the results.

        http://www.phoronix-test-suite.com/ [phoronix-test-suite.com]

      • (Score: 2) by Woods on Monday May 05 2014, @02:23PM

        by Woods (2726) <woods12@gmail.com> on Monday May 05 2014, @02:23PM (#39797) Journal

        4 g ought to be enough for anybody.

    • (Score: 3, Insightful) by VLM on Monday May 05 2014, @12:47PM

      by VLM (445) Subscriber Badge on Monday May 05 2014, @12:47PM (#39750)

      You can fix the slow IO by reducing IO by increasing memory, so just yank that dude apart and insert more memory. Of course if you're yanking it apart its 10x easier to find a larger replacement drive than replacement memory, and usually 10x easier to install, which would make the memory upgrade not as required anymore. Although you'd be better off doing both. So given a pile of cash of size X, you'd be better off using it to upgrade the disk.

      One way to analyze a system is via equivalent time. Given no specific data and being too lazy to bother guessing, making your IO slower doesn't make it unusable, it just makes it perform at X-5 mfgr year rather than X mfgr year. If you are "doing stuff" that would work great on a X-6 year old computer then you won. If on a X-4 year old computer then you lost. Although what it means to win or lose is very unclear.

      Another way to analyze hobby systems is by pain. What will make you hurt more, using the older SW (probably security holes and never updated and boring by now) or having a somewhat slower IO chain (unless you're doing real time stuff like watching video or burning DVDs, who cares?).

      • (Score: 4, Funny) by mrclisdue on Monday May 05 2014, @02:45PM

        by mrclisdue (680) on Monday May 05 2014, @02:45PM (#39806)

        My brain hurts a bit from reading your post, but had you compressed your thoughts my brain may have hurt less, or more. Depends on me.

        Had you expanded or expounded your thoughts, my brain may have hurt more, or less. Or I may not have bothered to digest it. I'm definitely older hardware, if that counts for anything.

        One way or another, it's win-win, I think.

        cheers,

        • (Score: 2) by VLM on Monday May 05 2014, @04:26PM

          by VLM (445) Subscriber Badge on Monday May 05 2014, @04:26PM (#39832)

          I believe a somewhat superior analogy would be:

          Drugs expand your mind although they make your body less productive, which really pisses off the people who feel entitled to tell you what to do with your body. And if you're going to the effort of obtaining and using psychedelics, you should get and use the "good" ones not the "bad" ones.

          The second point is something along the lines of being "still stuck in the 60s" isn't necessarily an automatic fail, if you liked the 60s.

          The last point is some (legal/bad) drugs mess with your digestion and give you hangovers, and personal tolerance to those effect varies widely.

          That's my commentary vs the rather simple "just say no" analogy.

          I think this discussion via analogies is pretty funny. The sex ed tangent could be even more hilarious "So... it doesn't fit, and various crackpots on the internet are offering free advice, ranging from organic soy oil to superglue, and ..."

    • (Score: 2) by mcgrew on Monday May 05 2014, @01:21PM

      by mcgrew (701) <publish@mcgrewbooks.com> on Monday May 05 2014, @01:21PM (#39762) Homepage Journal

      You could partition the drive into two logical drives, and compress one of them. Use the uncompressed HDa for apps and the compressed HDb for data. Some data (video files, for instance) may have to be on the uncompressed drive, butit would still free up space.

      --
      mcgrewbooks.com mcgrew.info nooze.org
    • (Score: 2, Interesting) by Jtmach on Monday May 05 2014, @02:35PM

      by Jtmach (1481) on Monday May 05 2014, @02:35PM (#39802)

      I've read several arguments that a decently fast CPU and a compressed filesystem, will result in marginally faster system performance overall.
      The thought being, if you are IO bound, you can send more compressed data across the pipe then uncompressed. Uncompressing the files is a relatively simple task for the CPU, so you get your data to the CPU a little faster, and the CPU does a little more work.
      In my tests, it didn't seem to make a difference either way, but I didn't spend a lot of time looking into it. In my particular case, it turned out to be a better use of my time to refactor the code.

    • (Score: 3, Insightful) by morgauxo on Monday May 05 2014, @02:56PM

      by morgauxo (2082) on Monday May 05 2014, @02:56PM (#39808)

      >>The main problem with laptop performance...

      Does performance always really matter? I used old laptops myself all through college and that was a decade ago so those laptops would be really old by today's standards. But, studying for a CS degree what did I need? The ability to compile some very short programs, search the web for reference material that was mainly text, occasionaly write a paper on a word processor...

      Depending what his needs are even an old netbook further cripled by compression overhead might be just fine! Maybe he wants to do more than that though. If he is only doing those kind of things I'm not sure what he needs the space for. Then again depending on what distro he is installing maybe he needs the space just to install all the default applications that he will never use!

      >>...which will drain your battery quicker

      Another good point. Back to my own experience though.. I used those old laptops because I could afford them. Replacement batteries are expensive, and even though I could have bought the batteries I wouldn't have just to use in what I knew was already way outdated, low value hardware. But that didn't make the laptops useless. I plugged in at the university library. These days there are plugs for people to use laptops in cafe's, restaurants and all sorts of places. Low battery life doesn't necessarily make his netbook useless.

      • (Score: 2) by The Mighty Buzzard on Monday May 05 2014, @03:16PM

        It entirely depends on what you use it for. I have an old pentium 133MHz laptop that is perfectly acceptable for reading ebooks on but on-the-fly compression would not be incredibly enjoyable to attempt on it. Now on a multi-core 1.5+GHz system that always/mostly runs on AC you might very well not notice any slowdown at all and battery life wouldn't be an issue.
        --
        My rights don't end where your fear begins.
  • (Score: 1, Insightful) by Anonymous Coward on Monday May 05 2014, @12:34PM

    by Anonymous Coward on Monday May 05 2014, @12:34PM (#39745)

    Can you try just the filesystem on a small portion of the drive to see first what the performance is?
    Otherwise, it sounds like you will be doing triple bypass surgery on an 80 year old.

    • (Score: 2, Interesting) by mcgrew on Monday May 05 2014, @01:24PM

      by mcgrew (701) <publish@mcgrewbooks.com> on Monday May 05 2014, @01:24PM (#39765) Homepage Journal

      Otherwise, it sounds like you will be doing triple bypass surgery on an 80 year old.

      Get off my lawn, you god damned asshole kid. If I need bypass surgery in twenty tears by God I'll have bypass surgery. Again, fuck you, you ageist asshole punk.

      --
      mcgrewbooks.com mcgrew.info nooze.org
      • (Score: 3, Informative) by mmcmonster on Monday May 05 2014, @01:45PM

        by mmcmonster (401) on Monday May 05 2014, @01:45PM (#39772)

        Considering the age of CABG patients has risen dramatically in the last 20 years ( http://www.ncbi.nlm.nih.gov/pmc/articles/PMC259997 8/ [nih.gov] ), you're probably in good company.

        • (Score: 2) by mcgrew on Monday May 05 2014, @05:45PM

          by mcgrew (701) <publish@mcgrewbooks.com> on Monday May 05 2014, @05:45PM (#39869) Homepage Journal

          Interesting, I hadn't seen that. I'll have to click the link to see if they know why (my guess is fewer people smoking). I doubt I'll need bypass surgery, though, as I don't have a single blood relative who ever had heart disease. I'll probably have lung problems instead, I smoked cigarettes for thirty years (quit in 2000) and pot for 40 (I still smoke pot, which has been shown to actually prevent cancers in cigarette smokers but can cause emphysema.

          They had health screenings at work last year before I retired, and said I was in good shape for a 40 year old and excellent for someone my age, so I'm not too worried. I'm more likely to be shot walking through the ghetto (I like to walk and don't always pay attention to my surroundings).

          --
          mcgrewbooks.com mcgrew.info nooze.org
      • (Score: 2) by fliptop on Monday May 05 2014, @03:21PM

        by fliptop (1666) on Monday May 05 2014, @03:21PM (#39817) Journal

        If I need bypass surgery in twenty tears by God I'll have bypass surgery

        Must..resist...urge...to...feed...troll...by...m aking...snarky...comment...about...Obamacare.

        --
        Our Constitution was made only for a moral and religious people. It is wholly inadequate to the government of any other.
        • (Score: 2) by mcgrew on Monday May 05 2014, @05:36PM

          by mcgrew (701) <publish@mcgrewbooks.com> on Monday May 05 2014, @05:36PM (#39864) Homepage Journal

          You're calling ME a troll? Your comment was abusive flamebait, and the reference to Obamacare was trolling at its finest, or you're too stupid to realize that Obamacare isn't used by those over 65, they have Medicare. Dumbass.

          Eighty may be past the age of death in your family, but not mine. At age 85 my mother, who is the baby of the family, is healthier than a lot of 40 year olds I know, and I had two friends die of heart disease at 40. Hell, she goes bowling every week. All but three of Mom's 12 older siblings are alive. Uncle Joe is in his late 90s and Mom says he looks more like 70.

          My paternal grandmother lived a hundred years. Her brother started smoking at age 12, quit at age 82, and died at age 94, Grandpa was killed in an industrial accident, victim of the Purina Corporation's greed. As I said, Dad died Friday you insensitive asshole. He'd been battling liver cancer for 2 years; the cancer was caused by his job. As an electrical lineman he was heavily exposed to PCBs and dioxins every day for 40 years until he retired 20 years ago. All of his co-workers died from the same disease decades ago. Yet you would have denied him care.

          Now fuck off and die, boy.

          --
          mcgrewbooks.com mcgrew.info nooze.org
          • (Score: 0) by Anonymous Coward on Tuesday May 06 2014, @04:08AM

            by Anonymous Coward on Tuesday May 06 2014, @04:08AM (#40017)

            My condolences.

            Anonymous because it's off-topic and you don't know me anyway.

    • (Score: 2) by ls671 on Monday May 05 2014, @04:14PM

      by ls671 (891) Subscriber Badge on Monday May 05 2014, @04:14PM (#39827) Homepage

      I just had quintuple bypass surgery and I am 86.

      --
      Everything I write is lies, including this sentence.
  • (Score: 4, Insightful) by Marand on Monday May 05 2014, @12:53PM

    by Marand (1081) on Monday May 05 2014, @12:53PM (#39756) Journal

    If the netbook has a spare USB port and your space is really that limited, maybe you'd be better served with a low-profile usb drive like this 32GB one [newegg.com] instead?

    Not sure if the I/O rate would be better or worse than trying to do filesystem compression, though.

  • (Score: 4, Funny) by wonkey_monkey on Monday May 05 2014, @12:55PM

    by wonkey_monkey (279) on Monday May 05 2014, @12:55PM (#39757) Homepage

    ...just won't die

    Famous last words...

    --
    systemd is Roko's Basilisk
  • (Score: 4, Insightful) by gringer on Monday May 05 2014, @01:29PM

    by gringer (962) on Monday May 05 2014, @01:29PM (#39766)

    Just replace the hard drive with a SSD that has (at least) double the capacity. You'll be thankful that you made the change.

    --
    Ask me about Sequencing DNA in front of Linus Torvalds [youtube.com]
  • (Score: 3, Interesting) by mmcmonster on Monday May 05 2014, @01:49PM

    by mmcmonster (401) on Monday May 05 2014, @01:49PM (#39776)

    I had a disk error recently and am looking to go to BTRFS for my next OS install.

    Anyone using it in general? Can I use it for / and /boot? Any caveats?

    Are there good drivers for OS X if I decide to share an external drive?

    Any issues with sharing a SMB folder?

    (I've been using ext4 since it's been solid and ext3 before that.)

    • (Score: 0) by Anonymous Coward on Monday May 05 2014, @02:30PM

      by Anonymous Coward on Monday May 05 2014, @02:30PM (#39801)

      I'd advise you to spend a few days lurking in #btrfs on Freenode, just to see what comes up. But my impression of it so far:
      - the filesystem and its subvolumes can be quite stable under normal operation
      - the newer and nicer features (subvolume quota, compression) can be oops-inducing.
      - some workloads (most notably, kmail indexing with Linux 3.13) can trigger filesystem corruption during normal operation
      - most importantly, its only out-of-space handling is to fail hard

      I'm using it as my backup target only, with disks that are much larger than the dataset. I don't trust it with live data yet.

      • (Score: 2) by mmcmonster on Monday May 05 2014, @03:04PM

        by mmcmonster (401) on Monday May 05 2014, @03:04PM (#39813)

        Considering I ran out of space twice on / in the last year, maybe I'll stick with ext4 for now. ;-)

        (Wasn't planning on running out of space on /. It just kind of happened.)

        • (Score: 2) by FatPhil on Monday May 05 2014, @07:37PM

          by FatPhil (863) <pc-soylentNO@SPAMasdf.fi> on Monday May 05 2014, @07:37PM (#39912) Homepage
          > Wasn't planning on running out of space on /. It just kind of happened.

          Wasn't planning on running out of patience on /. it just kind of happened.
          --
          Great minds discuss ideas; average minds discuss events; small minds discuss people; the smallest discuss themselves
      • (Score: 2) by sjames on Tuesday May 06 2014, @03:45AM

        by sjames (2882) on Tuesday May 06 2014, @03:45AM (#40013) Journal

        When the dust settles, it looks like btrfs will be more versatile than ZFS, particularly WRT management of RAID. However, for now btrfs shows it's immaturity.

        I recently tested it in a mirror setup on a throwaway VM. The biggest problem is that it seemed to completely fail to notice when I pulled a disk out from under it. I couldn't even inform it that the disk was gone. To be fair, the RAID stuff is new, and wasn't claimed to be production ready. That didn't alter the fact that I needed it working.

        For now, ZFS may be a better bet, but I wouldn't use it as my root filesystem. Perhaps have a seperate / and make the rest of the disk a zpool. Create home usr and var out of the pool.

        So for now I use ZFS but I intend to revisit btrfs once it has a little more time to mature.

    • (Score: 2) by Maow on Monday May 05 2014, @08:27PM

      by Maow (8) on Monday May 05 2014, @08:27PM (#39929) Homepage

      Ars Technica had a terrific series by Jim Salter comparing ZFS & BTRFS with a fair bit of detail on each.

      I don't have the link handy, but it ought to be easy to find.

      Also, Jim's pretty active in their forum (Linux Kung Fu forum) whence the article seems to have arisen from.

      Check it out; I plan on giving it a shot soon myself, after reviewing the article on BTRFS.

    • (Score: 1) by Sceptic on Wednesday May 07 2014, @04:23AM

      by Sceptic (1684) on Wednesday May 07 2014, @04:23AM (#40422)

      I wouldn't install it as a root filesystem again - it doesn't have an online filesystem checker, so you can't boot with a read-only filesystem and fix minor problems when it stuffs up.

      That being said, I haven't lost any important data to filesystem corruption yet, and have been running on it for months.

  • (Score: 3, Insightful) by tangomargarine on Monday May 05 2014, @01:57PM

    by tangomargarine (667) on Monday May 05 2014, @01:57PM (#39783)

    Um...btrfs isn't the only Linux FS to support compression, is it? Each time I've tried btrfs it's eventually gone wonky on me.

    --
    "Is that really true?" "I just spent the last hour telling you to think for yourself! Didn't you hear anything I said?"
  • (Score: 2) by marcello_dl on Monday May 05 2014, @02:11PM

    by marcello_dl (2685) on Monday May 05 2014, @02:11PM (#39790)

    Instead of compressing the filesystem you could install a live distro on the hd as if it were some usb storage. So you'll have a compressed squashfs image and some files or folders used for persistence.
    I would also explore usb keys and sd cards to use for storage or booting.

    • (Score: 1, Insightful) by Anonymous Coward on Monday May 05 2014, @06:29PM

      by Anonymous Coward on Monday May 05 2014, @06:29PM (#39890)

      install a live distro on the hd as if it were some usb storage

      The name for that is "frugal install". [google.com]

      ...and doesn't "HD" stand for "high definition" these days?
      Even back in the day it stood for "high-density floppy", with HDD being the abbreviation for hard disk drive.

      -- gewg_

    • (Score: 2) by quitte on Monday May 05 2014, @06:36PM

      by quitte (306) on Monday May 05 2014, @06:36PM (#39893) Journal

      I second this. I played around with that some and in my experience at least with lzma compression , which has a high compression ratio but is fast at decompressing (not sure how xz compares), the lowered io load far outweighs the increased cpu load. It felt way snappier. This was on an aspire one with its crappy ssd and also on a pentium 3 class system.
      I'd actually love to see the live system helper tools improve such that instead of running classical installations it became viable to rebuild images instead of updating or installing packages.
      This would give the additional benefit of added trust in your own system. A simple image rebuild would give a fresh install and all the differences that went into the system would be represented in the human readable meta-data.
      Of course this can already be done right now, but the tools at the moment are more prepared to create bootable media instead of installing bootable images to the boot partition. It's simply not quite convenient enough.

      • (Score: 2) by marcello_dl on Monday May 05 2014, @09:11PM

        by marcello_dl (2685) on Monday May 05 2014, @09:11PM (#39941)
        I'd actually love to see the live system helper tools improve such that instead of running classical installations it became viable to rebuild images instead of updating or installing packages Check out debian live and porteus, they have two very different approaches, but in both it's pretty easy to know what is added to the installation, or what is persisted. I have a debian live wheezy xfce dark theme for office use that is shaping up pretty well.
        • (Score: 2) by quitte on Tuesday May 06 2014, @04:15PM

          by quitte (306) on Tuesday May 06 2014, @04:15PM (#40209) Journal

          Thanks. I didn't know about porteus. However I managed to not give slackware a try and am not planning to change that.

          I know about debian-live. However I didn't come up with a solution that allows putting additional live images with their respective bootloader entries to the local hard disk. At least none that is anywhere near as simple as make "install". Also the resulting system should contain everything to rebuild itself just as easy. Maybe I'll find the time to have another go at it by the end of summer. But unless I can get some easy study credits out of it I guess I'll postpone that idea indefinitely.

  • (Score: 4, Insightful) by Subsentient on Monday May 05 2014, @02:38PM

    by Subsentient (1111) on Monday May 05 2014, @02:38PM (#39803) Homepage Journal

    It has no compression but by god, is it the most stable filesystem this side of the multiverse. I have been using ext4 since it first came out, and I have never had an ext4 filesystem become too damaged to repair from a power loss, and never had one become unreadable from anything short of hardware failure or zeroing it out.
    I strongly recommend you stay with ext4. It's rock solid. For thumbdrives, you can create ext4 without a journal to lengthen your drive's life as good as ext2: mkfs.ext4 -O ^has_journal /dev/sdXn

    --
    "It is no measure of health to be well adjusted to a profoundly sick society." -Jiddu Krishnamurti
    • (Score: 5, Informative) by juggs on Monday May 05 2014, @04:06PM

      by juggs (63) on Monday May 05 2014, @04:06PM (#39826) Journal

      I must second that sentiment.

      One caveat though. If using Western Digital drives with ext4 either turn off journalling (that reduces the reliability of the filesystem in case of hard crashes or power outages) or disable the 'intelligent' power management on the drive using idle3-tools.

      The issue is that the drive automatically unmounts the heads after a few seconds of idle time to save power, then the ext journalling process kicks in and wakes up the drive again causing the heads to be mounted again. This cycle repeats over and over ad infinitum. A couple of my drives reached over 500K load/unload cycles before I spotted the problem (well on the way to the manufacturer's quoted average failure point).

      Load/unload cycle count can be viewed in the drive's SMART data using smartctl (provided in package smartmontools for those on a debian flavoured distro), or palimpsest for a GUI approach.

      Just a heads up and totally off topic as regards the discussion at hand I know :D

      • (Score: 2) by hankwang on Tuesday May 06 2014, @09:47AM

        by hankwang (100) on Tuesday May 06 2014, @09:47AM (#40079) Homepage

        If using Western Digital drives with ext4 either turn off journalling (that reduces the reliability of the filesystem in case of hard crashes or power outages) or disable the 'intelligent' power management

        YMMV for "turn of journalling". A typical Linux installation has many processes that are touching files all the time, either by reading, thereby updating atime or by writing. A while ago, I tried setting up an old netbook as a NAS, but I couldn't get it to spin down the hard disk reliably while idle. IIRC, processes like samba and postfix were culprits.

        On a laptop with things like Firefox running, it also seems impossible to get the hard disk to spin down for power saving. And I'm not going to give up journalling.

        • (Score: 3, Interesting) by juggs on Thursday May 08 2014, @05:35AM

          by juggs (63) on Thursday May 08 2014, @05:35AM (#40820) Journal

          Hence why I mentioned disabling the on drive power management to prevent the heads unnmounting - re-mounting continually as the alternative. The heads up was for someone who puts up a default ext4 debian / ubuntu / centos install with "power saving" WD green, blue or red drives. In that scenario the drives will die prematurely due to the excessive head ramp on and offs, purely as a coincidence of timing between the drive powering down and the FS journalling process kicking in. It's better to keep the heads flying than demount them for a few seconds every few minutes (which is what happens in practice).

          Different considerations for a laptop with a full blown desktop GUI running - there's running applications wanting to write their current state every few seconds etc. on top of the underlying journalling in the FS (which with ext4 seems a litle hyperactive IMHO) - there's no realistic way to get the drive to spin down - you'd be better served going to look for lost pennies on the pavement as that would be more profitable than the power savings of spinning down a laptop drive now and then.

          Anyways, my original post was just meant as a warning for WD drive users - in their default shipped configuration they will ramp the heads on and off the platters every few seconds with ext journalling active - just a timing thing. Looking at the kit before me now, I see solely ext4 partitions and WD drives some quite geriatric, I'm not hating on WD by any means, just this particular power saving feature causes an issue in certain circumstances - for which I linked to a tool to remedy.

  • (Score: 2) by morgauxo on Monday May 05 2014, @02:58PM

    by morgauxo (2082) on Monday May 05 2014, @02:58PM (#39811)

    Just do it. If you don't like it then undo it. The problem is we all have different use cases as well as different tollerances for slowness and different financial means to buy something better. We can't tell you if this solution is right for you or not. Many THINK they know what you need. But only you can tell for sure. Just try it, you don't have anything to lose!

  • (Score: 2) by caseih on Monday May 05 2014, @03:37PM

    by caseih (2744) on Monday May 05 2014, @03:37PM (#39822)

    The lower end (and cheaper) Sandforce-based SSDs already do compression at a low level automatically. So on those drives, doing file system compression is not going to buy you anything in terms of speed or capacity.

    • (Score: 2) by ls671 on Monday May 05 2014, @04:23PM

      by ls671 (891) Subscriber Badge on Monday May 05 2014, @04:23PM (#39828) Homepage

      "The lower end (and cheaper) Sandforce-based SSDs already do compression at a low level automatically."

      Interesting. But how do they then specify the capacity of the drives since it is now going to be variable depending on file content?

      --
      Everything I write is lies, including this sentence.
    • (Score: 2) by ls671 on Monday May 05 2014, @04:38PM

      by ls671 (891) Subscriber Badge on Monday May 05 2014, @04:38PM (#39835) Homepage

      I just found this:

      It seems your argument isn't valid since the space saved by the SSD hardware compression isn't made available to the OS. So, compressing with BTRFS would still give him more space even in the setup you suggest.

      http://www.tomshardware.com/reviews/vertex-3-sandf orce-ssd,2869-3.html [tomshardware.com]

      "If the controller is compressing data, you might wonder why a SandForce-based drive can't just double its storage capacity. Say you have a 120 GB Vertex 2. If you filled up the drive with 50 GB of compressible data squished down into 25 GB of written information, where does the other 25 GB go? Well, it is added to the pool of overprovisioning. There is no way to use this as additional storage space because of the way an OS addresses LBA space. Operating systems recognize a static, not a dynamic, LBA. From a stability standpoint, you don't want the address of recently-written data to keep changing. However, there is a tangible benefit to compression besides writing less. Remember, more overprovisioning means less data has to be moved around as the drive is filled. This means you could actually get the benefit of enterprise level provisioning (28%) if you were writing easily-compressible data." ...

      --
      Everything I write is lies, including this sentence.
      • (Score: 2) by jimshatt on Monday May 05 2014, @07:45PM

        by jimshatt (978) on Monday May 05 2014, @07:45PM (#39915) Journal
        Compressing already compressed data doesn't work very well. This means that you would indeed have more space available to the OS, but none (or little) of the benefits of overprovisioning. That's probably okay though, in his case.
  • (Score: 0) by Anonymous Coward on Monday May 05 2014, @07:57PM

    by Anonymous Coward on Monday May 05 2014, @07:57PM (#39921)

    I've experienced two instances of total data loss with BTRFS on two out of three machines. The most recent was about 3 weeks ago on a new Slackware 14.1 install.

    Thankfully I had current backups and could just "rsync ...." the backup over and continue onward. Although when the Slack 14.1 BTRFS keeled over, the restore was onto ext4 instead.

    So, I'm down to only one BTRFS system now, and eventually when I upgrade it, it will be ext4 as well. The dire warnings of BTRFS not being ready for prime time, and to not trust irreplaceable data with it are all too true.

    So, feel free to try, but make sure you have good, current, backups.

  • (Score: 2) by Reziac on Tuesday May 06 2014, @04:10AM

    by Reziac (2489) on Tuesday May 06 2014, @04:10AM (#40019) Homepage

    ...you've been warned!!

    --
    And there is no Alkibiades to come back and save us from ourselves.