Announcement

Collapse
No announcement yet.

ZFS File-System Tests On The Linux 3.10 Kernel

Collapse
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • ZFS File-System Tests On The Linux 3.10 Kernel

    Phoronix: ZFS File-System Tests On The Linux 3.10 Kernel

    Using the latest ZFS On Linux support, the ZFS file-system was benchmarked from the Linux 3.10 stable kernel and compared to the Linux file-system competition...

    http://www.phoronix.com/vr.php?view=MTQyNzk

  • #2
    Back in like 2008 it was a big deal, now ZFS is pretty much irrelevant.

    Comment


    • #3
      Originally posted by phoronix View Post
      Phoronix: ZFS File-System Tests On The Linux 3.10 Kernel

      Using the latest ZFS On Linux support, the ZFS file-system was benchmarked from the Linux 3.10 stable kernel and compared to the Linux file-system competition...

      http://www.phoronix.com/vr.php?view=MTQyNzk
      it strikes me that these benchmarks look pretty synthetic, do you have any idea if there's some way to test something more akin to real usage performance? i've been playing with zfs and lz4 compression on ssd, and i compared zfs and md/ext4 for zfs, and i noticed that zfs cpu usage can be pretty high, and performance is generally lower, but in real use a lot of difference seems to disappear, and lz4 compression is nice

      Comment


      • #4
        Originally posted by mark45 View Post
        Back in like 2008 it was a big deal, now ZFS is pretty much irrelevant.
        btrfs is still buggy and runs out of disk space when there is still free disk easily on small partitions (like on ssd)

        lvm's snapshotting isn't great

        zfs is a bit "big", and a bit slow but it's stable and does tend to work pretty well.

        the zfs snapshotting and zfs send / zfs recv is great. it has lz4 compression as option now, it's having forward momentum.

        Comment


        • #5
          Originally posted by mark45 View Post
          Back in like 2008 it was a big deal, now ZFS is pretty much irrelevant.
          Let us know when there's something equivalent in Linux.

          Comment


          • #6
            I have been using ZFS(freebsd) on a few servers at home and work for about 2 years.
            I enjoy working with it and it has saved my butt a few times.

            Comment


            • #7
              Originally posted by johnc View Post
              Let us know when there's something equivalent in Linux.
              You mean like this ZFS on Linux test? I guess we should let you know now.

              Comment


              • #8
                Originally posted by smitty3268 View Post
                You mean like this ZFS on Linux test? I guess we should let you know now.
                Which exactly proves the point that ZFS is still most definitely relevant, since there are no viable competitors on Linux.

                Seriously, the Phoronix forums are worse than YouTube comments, which is pretty sad.

                Comment


                • #9
                  Originally posted by mercutio View Post
                  btrfs is still buggy and runs out of disk space when there is still free disk easily on small partitions (like on ssd)

                  lvm's snapshotting isn't great

                  zfs is a bit "big", and a bit slow but it's stable and does tend to work pretty well.

                  the zfs snapshotting and zfs send / zfs recv is great. it has lz4 compression as option now, it's having forward momentum.
                  The btrfs handling of free space is still a bit problematic (a bit, as in if you really work on it, you can still cause situation where you need to manually clean journal), but if you use mixed allocation trees then you can pretty much use the space up to last cluster for storing data. So if you really care for that, you can. Of course you will pay for that in performance (I have no idea by how much). For very small volumes mkfs.btrfs does that automatically anyway, so you should be OK with newly created file systems.

                  btrfs have had send/recv functionality for, I think, 2 kernel releases. lz4 shouldn't be too hard to add to btrfs too...

                  Originally posted by pdffs
                  Which exactly proves the point that ZFS is still most definitely relevant, since there are no viable competitors on Linux.

                  Seriously, the Phoronix forums are worse than YouTube comments, which is pretty sad.
                  Oh, Please! Shine on our unworthy minds, pdffs, You forever brilliant beacon of knowledge and enlightenment!

                  Comment


                  • #10
                    keep up the good work!

                    Originally posted by pdffs View Post
                    Which exactly proves the point that ZFS is still most definitely relevant, since there are no viable competitors on Linux.

                    Seriously, the Phoronix forums are worse than YouTube comments, which is pretty sad.
                    So, why don't you say something relevant and worthy. If you have nothing to say why not fuck off!

                    I believe phoronix has started not as that much expert in the field of benchmarking and reviewing at least like Anandtech. But they care enough to improve and it is one of the top 10 sites I wont miss reading their articles. I think with more help from the community, phoronix will become defacto standard of benchmarking/reviewing open source world.

                    Comment


                    • #11
                      Originally posted by mercutio View Post
                      lvm's snapshotting isn't great
                      What's wrong with it? Not trolling, just curious.

                      Comment


                      • #12
                        Btrfs may not be as polished yet, but as of the last few kernel releases it's pretty much ready to go IMO. Btrfs does have the advantage of superior handling of block devices and volumes IMO.

                        Comment


                        • #13
                          Originally posted by Chewi View Post
                          What's wrong with it? Not trolling, just curious.
                          If you have a snapshotted volume, every write to base volume is duplicated for every snapshot. So if you have 4 snapshots of the base volume, the disk sees 5 times as many writes. You can guess what that does to performance.

                          If you run out of space in the snapshot, the whole snapshot is lost forever, so if you actually want to keep it for longer time, you need to provision at least as much space as the base volume has.

                          (there were also other problems but I recall hearing that they have been fixed)

                          But with all the problems with them, I still use them, only for very specific tasks, under detailed supervision and so on. The btrfs snapshots are "plug and play" in comparison.

                          Comment


                          • #14
                            The situation isn't really ideal at the moment.

                            Ext4 is a great fast filesystem, however it is missing data integrity checks and the self healing mechanism of ZFS. I think this is a must have today.
                            ZFS is rock stable and brings all you want for a modern filesystem, but it is slow as hell on Linux and the development is pretty much dead.
                            BTRFS brings most of the features ZFS brings and some more I really like, like the automatic reallocation of hot data. However last time I used it, it seamed not ready for production.

                            I'm aware that what filesystem you use heavily depends on what use cases you have. I currently run ZFS on all my setups because stability and integrity are more important to me than performance. I would switch back to BTRFS if the problems I faced last time are fixed now. Have to make another test soon.

                            Comment


                            • #15
                              Originally posted by ZeroPointEnergy View Post
                              Ext4 is a great fast filesystem, however it is missing data integrity checks and the self healing mechanism of ZFS. I think this is a must have today.
                              Self-Healing like creating new? Or is this now fixed in ZFS?

                              ZFS is rock stable and brings all you want for a modern filesystem, but it is slow as hell on Linux and the development is pretty much dead.
                              BTRFS brings most of the features ZFS brings and some more I really like, like the automatic reallocation of hot data. However last time I used it, it seamed not ready for production.
                              A filesystem where you can't delete files if it is full, I wouldn't call rock stable. And this was observed on Solaris, not OpenSolaris or Linux. I'm amused about the ZFS-hype still around.

                              Comment

                              Working...
                              X