No announcement yet.

Btrfs Is Not Yet The Performance King

  • Filter
  • Time
  • Show
Clear All
new posts

  • Btrfs Is Not Yet The Performance King

    Phoronix: Btrfs Is Not Yet The Performance King

    With the release this week of Fedora 11 Preview, which incorporates install-time support for the Btrfs file-system into Red Hat's Anaconda installer, we have now delivered our first set of benchmark results for this next-generation Linux file-system. Through a horde of disk tests we have looked at the Btrfs file-system performance and compared it to that of EXT3, EXT4, and XFS. While Btrfs does perform well in some areas, it is not yet the performance king for Linux file-systems. As our results show, in some tests it even has a hard time competing with the incremental EXT4 file-system.

    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
    Why no Reiser4? It is the most similar to Btrfs as far as technology and is much more mature.


    • #3
      why no love for jfs here?

      I'd rather see a comparison against hammer than ext3/4 or reiser...


      • #4
        Btrfs is mainly created for the Oracle client that dont want to use "raw device"...
        It's to improve performance reading/writing large files with high concurrency.
        So it have to be fast on concurrent request.

        What should be looked is :
        how mysql perform on BTRFS
        how postgres perform on BTRFS
        how firebird perform on BTRFS

        As there is no magical solution, btrfs is no exception. It's not a general usage fs as are ext?.
        (imho) On desktop xfs will be the way to go.



        • #5
          Well, imho btrfs is not as interesting as everyone makes it out to be. Performance-wise it's obviously not so great (I do realise that it's still in development and this might change in the future), and the features it delivers are not very interesting as well imho, except maybe for the online defragmentation thingy. But I'm not an enterprise user whis is what this fs aims at I assume.

          Still I appreciate the work. Let's hope it doesn't get axed now that Oracle owns Sun and thus already has ZFS.


          • #6
            The Btrfs file-system supports writable snapshots, sub-volumes, object-level mirroring and stripping, data checksums, compression, online file-system checking support, and online defragmentation support.
            Does it have these features now? A few months ago, it didn't have defragmentation and other features were also missing.

            I hope the Oracle/Sun deal doesn't go through. I hate to say it, but I hope Microsoft steps in and saves the day...

            Oracle have done everything in their power to kill MySQL in the past. E.g. buying the ACID engine, and refusing MySQL to renew their license for ACID.

            So if the deal goes though, it is very likely the end of MySQL, Open Office and btrFS, as they now have ZFS.

            Let's hope that HP, Dell, or IBM buys Sun, and not Oracle or Microsoft.


            • #7
              btrfs - features... just needs to be competitive

              btrfs STOMPED in this early showing. Ext4 is a filesystem with OLD well known incremental features that SHOULD HAVE BEEN in ext3 (it's a basic filesystem with a few, very few, advancements).

              btrfs is feature rich... as long as it competes (and this EARLY, and I do mean, EARLY, test show that) it's ok.

              Winner: btrfs

              If you want a performance showdown, there are certainly scenarios that will work for this or that filesystem...

              I really do not care. People are 100% happy with ext3 performance (believe it or not).... I just hate the limitations of the filesystem... ext4 still has several of those limitations... btrfs won't have those limitations.


              • #8
                Seriously, why do you even bother testing benchmarks that are not even IO bound? Also, What about compression? Oh, that's right, it's a feature that those other filesystems don't have, and if you did test using it, you'd probably see a pretty decent bandwidth improvement, especially if done on easily compressable data like executables.


                • #9
                  What i would have liked to see on a filesystem benchmark is "updatedb"
                  That should probably show a clear difference between the filesystems.

                  Please consider adding updatedb to the set of tests in the suite.


                  • #10
                    looks like the better filesystem lacks much thats better.

                    I do however belive that a large mysql test would be good.