Syntax option:

    CHECKPOINT INTERVAL IS n BLOCKS

    You can limit how many transactions the database recovery process
    (DBR) must redo by setting a checkpoint interval. Setting a
    checkpoint interval instructs Oracle Rdb to periodically write
    modified pages to disk. This shortens recovery time.

    The value you assign to the checkpoint interval specifies the
    number of blocks the .aij file is allowed to increase to before
    updated pages are transferred. For example, if you set the
    checkpoint interval value equal to 100, all processes transfer
    updated pages to the disk when 100 blocks were written to
    the .aij file since the last checkpoint. Thus all processes
    contribute to .aij growth.

    If no checkpoint interval is established and a process completes
    1000 transactions but fails during number 1001, the DBR must redo
    transactions 1 through 1000 and undo number 1001.

    When a process attaches to the database, it writes a checkpoint
    record to the .aij file and notes the virtual block number (VBN)
    of the .aij file at which the checkpoint record is located. If
    the checkpoint is located at VBN 120 and the checkpoint interval
    is 100 blocks, the process checkpoints again when VBN 220 is
    reached.

    Because all processes contribute to .aij file growth, a process
    may be able to commit many transactions before checkpointing
    if update activity by other processes is low. Conversely, if a
    process' first transaction is long and if update activity by
    other processes is high, the process may be forced to checkpoint
    when it commits its first transaction.

    When the database checkpoint interval value is reached, Oracle
    Rdb executes the following steps:

    1. Writes updated pages to the disk.

    2. Writes a checkpoint record to the .aij file.

    3. Updates the run-time user process block (RTUPB) for each
       process to indicate where the checkpoint record is stored
       in the .aij file.

       The RTUPB is a data structure in the database root file that
       maintains information on each process accessing the database.
       The database recovery process (DBR) uses the RTUPB checkpoint
       entry to determine where in the .aij file recovery must start.