Announcement

Collapse
No announcement yet.

An Exciting Btrfs Update With Encoded I/O, Fsync Performance Improvements

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

  • An Exciting Btrfs Update With Encoded I/O, Fsync Performance Improvements

    Phoronix: An Exciting Btrfs Update With Encoded I/O, Fsync Performance Improvements

    SUSE's David Sterba on Monday submitted the Btrfs file-system updates for the in-development Linux 5.18 kernel...

    Phoronix, Linux Hardware Reviews, Linux hardware benchmarks, Linux server benchmarks, Linux benchmarking, Desktop Linux, Linux performance, Open Source graphics, Linux How To, Ubuntu benchmarks, Ubuntu hardware, Phoronix Test Suite

  • #2
    At this point is btrfs really next-gen and not just current-gen?

    Comment


    • #3
      Originally posted by geearf View Post
      At this point is btrfs really next-gen and not just current-gen?
      There is nothing "nexter" gen than btfs, while many mainstream things (ext4, ntfs in particular) are clearly lesser gen; hence, it is pretty adequate IMHO. Progress in file systems is somewhat slower than in other domains, due to obvious stability concerns and garbage hardware and software abstractions.

      Comment


      • #4
        Originally posted by mb_q View Post

        There is nothing "nexter" gen than btfs, while many mainstream things (ext4, ntfs in particular) are clearly lesser gen; hence, it is pretty adequate IMHO. Progress in file systems is somewhat slower than in other domains, due to obvious stability concerns and garbage hardware and software abstractions.
        Except the elephant in the room: OpenZFS

        Comment


        • #5
          If only they would upgrade the Zstd code too to the upstream version, everything would be perfect!

          Comment


          • #6
            allow reflinks/deduplication from two different mounts of the same filesystem
            This may be useful. My art site has original-sized files and reduced-size files in two locations, on two separate mounts, but sometimes they're the same file. We have the hashes to determine this, but the filesystem root is outside of the website's filesystem tree, so it would be nice not to have to refer to it. Thanks, Josef!
            Last edited by GreenReaper; 22 March 2022, 09:57 AM.

            Comment


            • #7
              Originally posted by mb_q View Post

              There is nothing "nexter" gen than btfs, while many mainstream things (ext4, ntfs in particular) are clearly lesser gen; hence, it is pretty adequate IMHO. Progress in file systems is somewhat slower than in other domains, due to obvious stability concerns and garbage hardware and software abstractions.
              It's next gen I guess still in how easy it makes to manage disks.

              Want to add a disk to a pool, change raid value, compress, decompress, snapshot the btrfs tool deserves a fucking award for how easy it is to use.

              Compared to LVM and ZFS its bliss.

              Comment


              • #8
                Originally posted by atomsymbol

                Question: Do you have personal experience with both Btrfs and OpenZFS on your machine(s)?
                Yes. My current ZFS zraid started off as a 1.5TB partition on a single 2TB HDD. The other 500GB was NTFS for Windows. When I quit dual booting around 2017 I gave the rest of the disk to ZFS. A few years later I bought a 4TB HDD and upgraded that to a mirror. A little bit later I bought a 2nd 4TB HDD and replaced my starting 2TB disk and finally had a 4TB mirror. Around that same time I bought a 4TB HDD for Windows because I was dual booting again. About a month or two ago I juggled data around and converted that mirror into a raidz with that 4TB HDD Windows was using (that's not an officially supported method and very dangerous).

                As far as being rock solid and reliable for my data, I've been a happy camper using ZFS since around 2016. The only time I have issues is when the kernel upgrades faster than OpenZFS. My remedy is to run custom kernels and to simply not update until OpenZFS does -- can't accidentally update to an incompatible kernel if Pacman doesn't know where to get it. I'm currently doing that with Linux 5.16 and 5.17.

                BTRFS I've always used for root and have been hit by GRUB and compression issues in the past. Sometimes it's from using esoteric setups and other times it's simply due to using a rolling release distribution and finding out a bug exists the hard way. Whatever the case may be, it has happened enough in the past 10 years that I've gone back to using Ext4 for my root volumes. Ext4 just works and I've never been hit by an issue so bad that a reinstall of the OS seemed like the easier fix.

                If I was using a distribution like OpenSUSE I'd probably have much better BTRFS experiences.

                Comment


                • #9
                  Originally posted by skeevy420 View Post

                  Except the elephant in the room: OpenZFS
                  How exactly is ZFS "next-gen" compared to btrfs, or was it just generic fanboy commentary?

                  Comment


                  • #10
                    Originally posted by skeevy420 View Post

                    Except the elephant in the room: OpenZFS
                    OpenZFS isn't in the kernel tree, nor is it license-compatible with Linux.

                    Comment

                    Working...
                    X