Announcement

Collapse
No announcement yet.

Testing The "Pretty Beefy" Btrfs Changes In Linux 3.2

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

  • Testing The "Pretty Beefy" Btrfs Changes In Linux 3.2

    Phoronix: Testing The "Pretty Beefy" Btrfs Changes In Linux 3.2

    Among many other enhancements and alterations, the Linux 3.2 kernel, the Btrfs file-system has some "pretty beefy" changes. Btrfs in Linux 3.2 merges in some long-standing Btrfs branches with new capabilities.

    http://www.phoronix.com/vr.php?view=16773

  • #2
    I wish it was finally marked as "stable".
    And: Is there any benchmark to test reliability of a FS?

    Comment


    • #3
      Its quite sad to watch btrfs progressing.
      It has been in development for a few years now, however still even lacks a consistency checker, shows often suboptimal performance in real-world use and tends to degrade over time.
      I lost data two times, because a btrfs-partition couldn't be mounted from one day to another.

      Guess it will take another 2-3 years until btrfs is ready to replace ext4.

      Comment


      • #4
        Originally posted by Linuxhippy View Post
        It has been in development for a few years now, however still even lacks a consistency checker, shows often suboptimal performance in real-world use and tends to degrade over time.
        As far as I understand, they have actively refused to deliver inconsistent filesystem checker for sole reason of not wanting that bad first impression.
        This way or another, ext4, xfs or even raiser3 are still options to utilize depending on FS task.
        Its pretty bad Oracle refused to release ZFS under more liberal license, same reason why OpenOffice got forked, I guess thats because a lot of people do not appreciate Oracle.
        It would be no wonder, if ZFS will be pushed under BSD or GPL(less possible) the day BTRFS gets stable.

        Comment


        • #5
          Originally posted by crazycheese View Post
          As far as I understand, they have actively refused to deliver inconsistent filesystem checker for sole reason of not wanting that bad first impression.
          This way or another, ext4, xfs or even raiser3 are still options to utilize depending on FS task.
          Its pretty bad Oracle refused to release ZFS under more liberal license, same reason why OpenOffice got forked, I guess thats because a lot of people do not appreciate Oracle.
          It would be no wonder, if ZFS will be pushed under BSD or GPL(less possible) the day BTRFS gets stable.
          Except that there is no point for Oracle in doing that. They are the main developers of both ZFS and BTRFS.

          Comment


          • #6
            Originally posted by Adarion View Post
            I wish it was finally marked as "stable".
            And: Is there any benchmark to test reliability of a FS?
            The "let it be used for 5 years in the enterprise, watch news headlines" bench is a pretty good one. If you need faster results, start a long disk test and periodically yank out the power cord.

            Comment


            • #7
              Hehe, you're probably right there. But pulling the power cord or sudden power loss most times resulted in drama here.

              Comment


              • #8
                "Rome wasn't built in a day"

                Seriously, I use XFS, XFS has been around for over 16 years now, they're still improving it and fixing bugs.

                BtrFS has been around for all of like a few years now, it's still incredibly new, some things it does have never even been done before. Back off and quit complaining that it's "Not done yet" every few days. Please?

                It'll be ready when it is ready, in the mean time be thankful that any Linux file system beats what Microsoft and Apple push.

                Comment


                • #9
                  That's something that's always interested me, you only see the Linux filesystems compared against themselves.
                  How do they stack up against NTFS?
                  With the NTFS module in the kernel, it would be possible to add comparative benchmarks to this wouldn't it?

                  Comment


                  • #10
                    Originally posted by RoboJ1M View Post
                    That's something that's always interested me, you only see the Linux filesystems compared against themselves.
                    How do they stack up against NTFS?
                    With the NTFS module in the kernel, it would be possible to add comparative benchmarks to this wouldn't it?
                    Who needs NTFS?!
                    For normal use there is ntfs3g and ntfs in-kernel drivers. For both the performance is just adequate to drag and drop stuff for compatibility.
                    If you think about using NTFS professionally, you should first clear its legal status; then you will probably be looking into commercial driver. Defragmentation and inconsistency issues included.

                    Comment


                    • #11
                      I don't need NTFS.
                      I just would like to know how the linux filesystems stack up against NTFS.
                      Is NTFS slow? Fast? The same?
                      It's something that's always just interested me.

                      Comment


                      • #12
                        The in-kernel NTFS driver is too incomplete to even run most FS benchmarks, and comparing a kernel driver to a FUSE driver isn't really comparing the filesystems. That's probably why you don't see too many comparisons of the kind you're asking about.

                        Comment


                        • #13
                          Originally posted by Ex-Cyber View Post
                          The in-kernel NTFS driver is too incomplete to even run most FS benchmarks, and comparing a kernel driver to a FUSE driver isn't really comparing the filesystems. That's probably why you don't see too many comparisons of the kind you're asking about.
                          NTFS is bloated, poorly designed, only has POSIX file system attributes as an afterthought to comply with the minimal requirements of the US government. It has higher overhead, slower performance, and much more upkeep than a typical Linux/UNIX file system.

                          Almost any file system you're likely to use on Linux or the BSDs is an improvement. I personally prefer XFS, on recent kernels the performance is excellent and so is the reliability and overall robustness.

                          NTFS is also heavily patented by Microsoft, which is why the in-kernel driver is in the shape it is. They have to watch out what they put in it or it's another thing Microsoft will claim is an infringement. There's a much more complete userspace module for NTFS called NTFS-3G which is much better if you need to work on Windows file systems from a Linux system.

                          Comment

                          Working...
                          X