I am having doubts about user-managed hot backups. Why does Log Writer write block images of changed blocks of the datafile in backup mode to the redo log instead of just the row information? Ultimately during recovery all the archived redo log files are applied to the restored files (after its log sequence number at the time of backup). Can you please explain this? Thanks.

    Requires Free Membership to View

    Login

When you are copying an active datafile, the datafile contents may be changed while you are copying the data to your backup location. This means that your backup will be "inconsistent." The last part of the backup may contain newer transactions and the first part may contain older transactions. To solve this problem, you must be running in Archive Log mode so that Oracle can recover any transactions made while you were performing your backup.

While copying a block of a datafile, it is possible that the block may be inconsistent. This is called a "fractured block." It is possible that a row in that block may be inconsistent as well. Part of that row is old in your backup and part of the row is new in that backup. Oracle decided the best way to handle a fractured block is to write an image of that block to the archived redo log. On recovery, Oracle does not have to figure out which row is inconsistent since it replaces the entire block.

Dizwell Informatics has some great information on this as well.

This was first published in June 2006

Join the conversationComment

Share
Comments

    Results

    Contribute to the conversation

    All fields are required. Comments will appear at the bottom of the article.
    Back to top