Prior to
Oracle8, DBAs controlled checkpoint frequency by setting the initialization file
parameters LOG_CHECKPOINT_INTERVAL (number of redo blocks between
checkpoints) and LOG_CHECKPOINT_TIMEOUT (in seconds) and setting the size
of the redo log files. In addition, Oracle always performs a checkpoint whenever a
log file switch occurs.
Reducing the checkpoint interval or timeout would result in smaller amounts of data
between checkpoints and lead to faster recovery times, but could also introduce the
overhead of more frequent checkpoints and their associated disk activity. A common
strategy for minimizing the number of checkpoints was to set the initialization
file parameters so that checkpoints would occur only with log switches.
Oracle8i introduced an initialization file parameter to provide a simpler and more
accurate way to control recovery times: FAST_START_IO_TARGET. The bulk of
recovery activity involves performing I/O for reading database blocks into the cache
so that redo can be applied to them. This parameter set a target ceiling on how many
database blocks Oracle would have to read in applying redo information.
Pages:
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572