Results 1 to 7 of 7
  1. #1
    Join Date
    Oct 2003
    Posts
    4

    Question Unanswered: Log size impacting system performance

    Usually I always use big database logs (200M - 50000 pages) . This time, increasing log size from 17M to 200M caused database to slow down significantly.

    vmstat was showing lots of "b" processes

    After log size was reduced, database went back to normal.

    Can someone please comment why increasing log size cause Log I/O to conflict with the other database I/O?

    Notice - database located on local storage array, RAID-5, same volume used for datafiles and logs.

  2. #2
    Join Date
    May 2003
    Location
    USA
    Posts
    5,737
    I don't know what the performance problem is, but think you should try to move the logs to another array. You don't ever want to loose the logs and the data. I realize that it is Raid-5, but strange things can happen.

    I assume that you can satisfy the same objective by using the smaller log file size, but define more log files.

  3. #3
    Join Date
    Oct 2003
    Posts
    4
    Hello Marcus,

    The question is why reducing log size reduced Log I/O contention with other database I/O.

    Having logs and datafiles in different palces is not a question for me, however at this point is not an option.

    Also, preference is to have big logs - 17 M logs are switching almost every minute

  4. #4
    Join Date
    Jun 2003
    Location
    Toronto, Canada
    Posts
    5,516
    Provided Answers: 1
    Originally posted by rwarshaw
    Hello Marcus,

    The question is why reducing log size reduced Log I/O contention with other database I/O.

    Having logs and datafiles in different palces is not a question for me, however at this point is not an option.

    Also, preference is to have big logs - 17 M logs are switching almost every minute
    I'd say, it's the log buffer size that's affecting performance in your case, not the log file size.
    ---
    "It does not work" is not a valid problem statement.

  5. #5
    Join Date
    Oct 2003
    Posts
    4
    Can you please explain why and how it can be confirmed from the database statistics?
    Last edited by rwarshaw; 10-26-03 at 17:08.

  6. #6
    Join Date
    Aug 2001
    Location
    UK
    Posts
    4,650
    Have a look at the thread
    http://www.dbforums.com/showthread.p...light=LOGBUFSZ

    for information on how to identify log buffering using snapshot ...

    Cheers
    Sathyaram

    Originally posted by rwarshaw
    Can you please explain why and how it can be confirmed from the database statistics?
    Visit the new-look IDUG Website , register to gain access to the excellent content.

  7. #7
    Join Date
    Oct 2003
    Posts
    4
    Thanks for referring.

    Still, since this parameter was not changed, it does not explain, why I/O contention happened

Posting Permissions

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