Results 1 to 4 of 4
  1. #1
    Join Date
    Jan 2004
    Posts
    12

    Unanswered: Filesystems for DB2

    Does anyone have any test results that show if DB2(8.2.3) performs better with multiple filesystems under AIX 5.3 JFS2?

  2. #2
    Join Date
    May 2003
    Location
    USA
    Posts
    5,737
    What exactly do you mean by "multiple filesystems"?
    M. A. Feldman
    IBM Certified DBA on DB2 for Linux, UNIX, and Windows
    IBM Certified DBA on DB2 for z/OS and OS/390

  3. #3
    Join Date
    Jan 2004
    Posts
    12

    Multi filesystems

    I generally break it down to put the logs on a filesystem, syscat on a filesystem, and several filesystems for data and indexes. I use multiple containers with dms tablespaces so I can break up the data and indexes. I have test results from this type of setup on ZOS moving datasets so I don't have any hot spots. In ZOS, the hot spots happen because there is 1 UCB to each volume so if the hot datasets are on the same volume, the i/o queue time goes up. Does the same thing happen in AIX?

    Currently, my client's setup is 1 filesystem for each instance. All application data/idx, logs, syscat, sort are all on the same filesystem. I am not the strongest working with AIX, but I believe I am going to be able to run a few tests with both types of setups and get some good information. I figure will be able to see most of the info I need from snapshots, but I think iostat will give me much more information.

    I was hoping someone had some test results similar to this situation.

  4. #4
    Join Date
    May 2003
    Location
    USA
    Posts
    5,737
    Spreading the data out over multiple disks can help, but there are many variables involved in determining how much of a factor it will be. This includes the amount of memory allocated for bufferpools and other memory heaps (like package cache). Memory is a lot cheaper on AIX boxes than on mainframes, and for OLTP systems it is not unusual for bufferpool hit ratios to be so high that disk performance is irrelevant.

    On decision support systems there are many things that can be done in configuring DB2 for Linux, UNIX, and Windows to minimize disk contention and I/O costs.
    M. A. Feldman
    IBM Certified DBA on DB2 for Linux, UNIX, and Windows
    IBM Certified DBA on DB2 for z/OS and OS/390

Posting Permissions

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