Announcement

Collapse
No announcement yet.

Oracle Talks Up Btrfs Rather Than ZFS For Their Unbreakable Enterprise Kernel 6

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

  • #81
    Originally posted by lucrus View Post

    I've been reading about RADI1C3 and RAID1C4 in btrfs and that they are more or less equivalent to RAID-Z and RAID-Z2, though I can't really tell if it's true nor why. Anyway, assuming it's not true and thus assuming RAID-Z is superior, that leads me to think the only advantage of ZFS is on bare metal, while if I'm using say LXD in a KVM instance I can safely go with btrfs, right?
    No, if I understand the documents correctly, they aren't. RAID-Z and Z2 are more like improved versions from RAID-5/6.
    In terms of mere redundancy yes RAID1Cn are definitely superior, but come with much higher cost: instead of storing parities RAID1Cn stores n complete copies.
    RAID5/6 hits the sweet point when you're balancing redundancy and storage efficiency (even after including the ZFS tax).

    Originally posted by lucrus View Post
    I assume you mean they need RAID-Z, because some RAID is provided by btrfs too.
    The challenge is the only reliable RAID levels from btrfs are RAID0/1 (or 0+1).
    I'm using btrfs as JBOD now and it works fine, but I won't risk my data to test btrfs's RAID5.

    Originally posted by lucrus View Post
    doesn't that hold true for ZFS too?
    Yes, they're equally unattractive to common users.
    So I also don't quite understand why Canonical is pushing ZFS on Linux for desktop...

    Comment


    • #82
      Originally posted by lucrus View Post
      I've been reading about RADI1C3 and RAID1C4 in btrfs and that they are more or less equivalent to RAID-Z and RAID-Z2
      Wrong. They are "equivalent" only in terms of perceived reliability, and not in this order. raid1c3 can survive outage of any two disks, just like raidz2 or raid6, while raidz (or raid5) is "equivalent" to raid1.

      Similarities end here. With raidz or raid5, you need just one extra disk with parity data, whereas with raid1 you have to double each disk. Whereas with raid1c3... you have to triple each disk to make that guarantee.

      So, yeah, raid1/raid1c3 is equivalent to raid5/raid6 in the same sense as owning a bicycle is equivalent to owning a Tesla, implying that both devices can get you somewhere

      Comment


      • #83
        Originally posted by Eraserstp View Post
        For me, the main drawback of zfs was the inability to expand the array by adding a new disk.
        I don’t know if such a feature exists now, but I chose btrfs precisely because of the ease of expanding the array.

        Btrfs does work pretty well with a complete mis-mash of disks. While you can stripe zfs across individual disks, you loose all redundancy and failure of any disk will take out the whole pool. Though zfs is fairly easy to expand a mirror at a time.

        Really there's good points to either.

        Comment


        • #84
          Originally posted by jacob View Post

          I don't see that happening. Only Canonical is barracking for ZFS, no-one else in the Linux community seems to care much for it.
          There are enough people interested in using ZFS on Linux that the Linux port became the upstream project.

          Comment


          • #85
            Originally posted by aht0 View Post

            Might not be that easy. NetApp and Oracle had court case over ZFS once in the past. Well, they settled it but on what terms..?
            Might it be that Oracle is tied to an non-disclosed agreements with NetApp and that's the reason it's not touching ZFS license with 10-feet pole, nor would go ahead and port it to Linux.

            Oracle could afterall easily use ZFS (as similar out-of-tree port) on it's RHEL re-spin - it's not going to sue itself over copyrights!!! - and gain competitive advantage while suing everybody else who'd try to run ZFS on Linux.

            But in reality it's not doing anything with ZFS at all, excepting use on Solaris. Illogical - unless there are factors in play wider public knows nothing about.
            Yeah, to us this seems like an illogical move.. but I really think Oracle does not *want* ZFS on Linux. They want you to buy a ZFS Oracle Storage Appliance. ZFS on Linux/FreeBSD cuts into their market because people do run large storage arrays on ZFS. BTRFS does not as people don't use it that way.

            In other words.. Oracle wants the small fish to use btrfs, and the large fish to use the Oracle Storage Appliance. ZFS on Linux/FreeBSD is a threat to them.

            As you remember they fired Mark Maybee for talking about unifying the ZFS code base.
            https://twitter.com/bcantrill/status...351360?lang=en
            Last edited by k1e0x; 05-26-2020, 02:42 PM.

            Comment


            • #86
              Originally posted by k1e0x View Post

              Yeah, to us this seems like an illogical move.. but I really think Oracle does not *want* ZFS on Linux. They want you to buy a ZFS Oracle Storage Appliance. ZFS on Linux/FreeBSD cuts into their market because people do run large storage arrays on ZFS. BTRFS does not as people don't use it that way.

              In other words.. Oracle wants the small fish to use btrfs, and the large fish to use the Oracle Storage Appliance. ZFS on Linux/FreeBSD is a threat to them.

              As you remember they fired Mark Maybee for talking about unifying the ZFS code base.
              https://twitter.com/bcantrill/status...351360?lang=en
              Could be.

              Comment

              Working...
              X