3 Replies Latest reply on Apr 22, 2013 2:12 PM by Reidod

    zpool concat performance


      We have a zpool for a Oracle data area as below.

      Unfortunately we have very slow performance on the zpool (and very bad backup performance as well) so I was wondering if anyone had any thoughts about zpool/zfs performance on a concatenated volume.. Ive read the usual performance guides but a couple of forums said the performance could be worse with a concat volume especially on an intensive Production read/write volume.

      oradata 1.76T 272G 906 265 13.7M 3.47M
      c3t60060E800544C900000044C90000105Ad0 1.56T 270G 811 192 12.2M 3.10M
      c3t60060E800544C900000044C90000105Bd0 198G 1.30G 94 73 1.43M 377K

      for info, the server is an M5000 running Solaris 10.

        • 1. Re: zpool concat performance
          I don't understand what you mean by a concatenated volume. Are you referring to some underlying hardware RAID config?
          I see a non-redudant ZFS storage pool of 2 LUNs so data is striped across each LUN.

          There are several ways to improve the performance of an Oracle database:

          1. Hardware RAID or a mirrored ZFS storage pool is recommended.
          2. Which Solaris 10 release is this? The latest Solaris 10 release is recommended.
          3. How many db file systems and how are they configured. A separate (small) pool for the redo log is also recommended.

          You can read about these recommendations here:


          Thanks, Cindy
          • 2. Re: zpool concat performance
            Yes, as far as my understanding there is no concat / stripe in zpool, it takes all as a single disk once bring it under zpool control. check out the IO operations and if possible, try to install Oracle in a local disk and see a performance...to decide if there is a problem in LUN.
            • 3. Re: zpool concat performance

              It is not advisable to set up zfs pools or file systems on top of SVM volumes. This configuration can cause performance issues.