Announcement

Collapse
No announcement yet.

9-Way File-System Comparison With A SSD On The Linux 3.17 Kernel

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

  • #11
    Originally posted by liam View Post
    IMHO, btrfs offers enough features that it is with moving towards with the understanding that some features are better supported than others. It now seems solid enough that I'd recommend its usage for an average user as long as they are using conservative mount options.
    Even with those you get fantastic benefits you have an intruding enough fs that it is able to act as the foundation to lennart's new proposal regarding system rollouts/updates/versioning.
    Eventually, it should contain features that no other non-clustering fs offers.
    A file system does not need to offer a ton of features to me. I use as little as is necessary and expect it to give me just that while doing a good job at it. It mainly needs to work without me having to tweak it or to retweak it every now and then.

    I also do not want to switch to RedHat for all my needs. I do not want to limit myself by being dependant on a single distributor or just a single file system. This has never been a dream of mine.

    Comment


    • #12
      Originally posted by sdack View Post
      A file system does not need to offer a ton of features to me. I use as little as is necessary and expect it to give me just that while doing a good job at it. It mainly needs to work without me having to tweak it or to retweak it every now and then.

      I also do not want to switch to RedHat for all my needs. I do not want to limit myself by being dependant on a single distributor or just a single file system. This has never been a dream of mine.
      It's not just that it offers tons of features but that it enables usage that wasn't feasible before. If you install opensuse it should just work and not require tweaking.
      I don't understand the red hat comment. This is all true oss. RH is as oss fanatic as any org outside of the fsf. Not zealots, mind you, but completely cognizant of oss benefits, just like Linus.

      Comment


      • #13
        Originally posted by liam View Post
        It's not just that it offers tons of features but that it enables usage that wasn't feasible before. If you install opensuse it should just work and not require tweaking.
        I don't understand the red hat comment. This is all true oss. RH is as oss fanatic as any org outside of the fsf. Not zealots, mind you, but completely cognizant of oss benefits, just like Linus.
        You talk about features like anyone would need an inconceivable amount of them. I do not. A file system needs to work and does not need to be a swiss army knife. A swiss army knife is not very useful, because it is not very good at anything, but only offers a lot of tiny tools. It is however not a very good knife nor is it a very good screw driver nor are its scissors of much use. It is not necessary to have a large range of features within a single file system when there is a range of file systems, and which covers most of it. Their competition to one another guarantees that it will stay this way.

        If I need a file system for a new SSD then why use btrfs when F2FS is faster? And why use btrfs for my old disks when EXT4 is compatible to EXT2/3 and just as fast as btrfs? These are two features I already do not get with btrfs regardless of what else it offers.

        The RedHat comment is about their plans on creating a packaging system on top of btrfs, which means you need to have btrfs and it is just not a choice any longer, but you become dependant on one single file system. Such a dependency is not a good feature for me even if it is for the purpose of forcing me into a world of bliss like it Microsoft does. This is RedHat's plan for the future of their users and I wish to stay away from it.

        Comment


        • #14
          1. I dont agree to the conclusion of the article technicaly the sentences are true, but it implies that btrfs is slower then as example ext4 because it wins less tests, and u get in exchange for that some more features.

          But if you look through the the tests btrfs is 6 times of 9 faster than ext4. So for me it looks like btrfs is faster than ext4. thats what I see on this numbers (ok that is when u dont weighten them differently).

          2. some say they do not care about the more features of btrfs. But who really dont uses lvm or mdadm with their OS do you really use bare metal normal partitions without lvm or mdadm? I used for several years even before I heard of btrfs the first time lvm and mdadm. My standard would be to at least use lvm as base. I always wonderd how such technics form the 80s? like *rofl* dos partition talbe could live for so long. Today I use most of the time btrfs directly on the harddisk without partitions. And god I love that.

          So if you are not a small minority, that do not care about that comfort in changing sizes and have one fs to use the space of your ssd better with features like quota or having virtually another fs for home... as technical intusiast u would use lvn at least with the other fses anyway.

          So this tests say not to much, even if u ignore the mount options, u would have to compare lvm + fs vs btrfs. Else its like u compare MS Word with Linux.

          And espcialy then with the cpu time used... because I guess that would be the biggest point of interest, with a core i7 I guess lvm would not matter but with a Intel atom or even maybe with a core i3 or a not highend amd cpu it would matter.

          Comment


          • #15
            Originally posted by sdack View Post
            You talk about features like anyone would need an inconceivable amount of them. I do not. A file system needs to work and does not need to be a swiss army knife. A swiss army knife is not very useful, because it is not very good at anything, but only offers a lot of tiny tools. It is however not a very good knife nor is it a very good screw driver nor are its scissors of much use. It is not necessary to have a large range of features within a single file system when there is a range of file systems, and which covers most of it. Their competition to one another guarantees that it will stay this way.

            If I need a file system for a new SSD then why use btrfs when F2FS is faster? And why use btrfs for my old disks when EXT4 is compatible to EXT2/3 and just as fast as btrfs? These are two features I already do not get with btrfs regardless of what else it offers.

            The RedHat comment is about their plans on creating a packaging system on top of btrfs, which means you need to have btrfs and it is just not a choice any longer, but you become dependant on one single file system. Such a dependency is not a good feature for me even if it is for the purpose of forcing me into a world of bliss like it Microsoft does. This is RedHat's plan for the future of their users and I wish to stay away from it.
            Btrfs isn't a swiss army knife, though I can understand how one could see it that way. It actually is the exact opposite of a swiss army knife in that it either currently allows, or plans to enable, better/safer ways of dealing with data, but I won't bother you any further with this.
            If speed is your only concern for disks then there's little reason to move to btrfs unless you run massively concurrent loads, in which case btrfs or xfs might be the faster choices.
            Lennart isn't RH. If you read the mailing lists you'd know that. People who seriously liken RH to ms pick and choose their data points. RH will NOT force customers to use btrfs. Many customers use xfs because of its extreme scaling/threading abilities. RH knows it's customers and choice is a major part of the reason why they are used. Not the major reason, but it's one that RH acknowledges.

            Comment

            Working...
            X