Page 4 of 4 FirstFirst ... 234
Results 31 to 36 of 36

Thread: EXT4 File-System Looks To Do Well Against NTFS

  1. #31
    Join Date
    Mar 2009
    Location
    Hellas
    Posts
    1,113

    Default

    True, these tests are not completely necessary but if a system has problems in some of these, that could have an impact and in other applications as well.
    Personally, I wouldn't run a lot of these, but since PTS has the ability and it's easy to do so, Michael choosed to run them. We can't blame him for them, can we?

  2. #32
    Join Date
    May 2007
    Location
    Third Rock from the Sun
    Posts
    6,587

    Default

    Quote Originally Posted by gilboa View Post
    Here's where your argument completely fails.
    Out of the box performance may be relevant when you're talking about desktop applications (to some extent) such as games and general utilities (lzma/gzip, mp3 encoding, etc).
    However, out of the box experience loses all meaning once you start talking about server applications such as apache or postgresql and specialized applications such ray-tracing or simulation software.

    - Gilboa
    I wouldn't say they lose all meaning as many servers out there are running distro default kernels.

  3. #33
    Join Date
    Oct 2006
    Location
    Israel
    Posts
    626

    Default

    Quote Originally Posted by deanjo View Post
    I wouldn't say they lose all meaning as many servers out there are running distro default kernels.
    I believe you missed my point:
    I never claimed that PTS should use custom kernels - quite the contrary.
    I believe PTS should use the default distribution kernel and the default distribution <b>packages</b> (whenever available).
    I am claiming that it is wrong to use the default non-optimized -configuration- (such as ext4 mount options) when it comes to server / specialized applications.

    - Gilboa
    DEV: Intel S2600C0, 2xE52658V2, 32GB, 4x2TB, GTX780, F20/x86_64, Dell U2711.
    SRV: Intel S5520SC, 2xX5680, 36GB, 4x2TB, GTX550, F20/x86_64, Dell U2412..
    BACK: Tyan Tempest i5400XT, 2xE5335, 8GB, 3x1.5TB, 9800GTX, F20/x86-64.
    LAP: ASUS N56VJ, i7-3630QM, 16GB, 1TB, 635M, F20/x86_64.

  4. #34
    Join Date
    May 2007
    Location
    Third Rock from the Sun
    Posts
    6,587

    Default

    Quote Originally Posted by gilboa View Post
    I believe you missed my point:
    I never claimed that PTS should use custom kernels - quite the contrary.
    I believe PTS should use the default distribution kernel and the default distribution <b>packages</b> (whenever available).
    I am claiming that it is wrong to use the default non-optimized -configuration- (such as ext4 mount options) when it comes to server / specialized applications.

    - Gilboa
    That maybe true on not customizing the kernel for peak performance on servers but there are a lot more servers out there running the default kernels then customized ones so I wouldn't say that the tests are meaningless.

  5. #35
    Join Date
    Apr 2008
    Location
    Saskatchewan, Canada
    Posts
    466

    Default

    Quote Originally Posted by deanjo View Post
    That maybe true on not customizing the kernel for peak performance on servers but there are a lot more servers out there running the default kernels then customized ones so I wouldn't say that the tests are meaningless.
    You don't need to build a custom kernel to use non-standard mount options (e.g. noatime) and anyone who cares about disk performance on a server is likely to be running with some. So there are good arguments in both directions; benchmarking is hard .

    That said, I've found ext4 so unreliable that I would never trust on a server with data I care about; both of my PCs with ext4 partitions have had it fail so badly that they wouldn't even boot and required manual fsck... I've ended up configuring them with data journaling just to stop ext4 eating my files. Even btrfs, which is supposed to be 'experimental', is proving more reliable.

  6. #36
    Join Date
    Oct 2006
    Location
    Israel
    Posts
    626

    Default

    Quote Originally Posted by deanjo View Post
    That maybe true on not customizing the kernel for peak performance on servers but there are a lot more servers out there running the default kernels then customized ones so I wouldn't say that the tests are meaningless.
    My English may be failing me:
    Again: If you deploy 100 web servers or 100 DB servers you will most likely use an enterprise distribution (such as RHEL) and the -default- distribution kernel with no optimization what-so-ever.
    Why? Because you pay huge amounts of money on per-server licenses just to get support, and once you change one CONFIG_XXX within the kernel, you no longer support.
    However, if you IT manager worth 10% of his paycheck, he'll optimize the hell out of the machine by changing the machine -configuration-.
    E.g. application configuration (postgres.conf), kernel configuration (sysctl) and FS mount options.

    - Gilboa
    DEV: Intel S2600C0, 2xE52658V2, 32GB, 4x2TB, GTX780, F20/x86_64, Dell U2711.
    SRV: Intel S5520SC, 2xX5680, 36GB, 4x2TB, GTX550, F20/x86_64, Dell U2412..
    BACK: Tyan Tempest i5400XT, 2xE5335, 8GB, 3x1.5TB, 9800GTX, F20/x86-64.
    LAP: ASUS N56VJ, i7-3630QM, 16GB, 1TB, 635M, F20/x86_64.

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •