First page Back Continue Last page Graphics
MTTR Advisory
Companion to FAST_START_MTTR_TARGET
- Number of seconds to perform crash recovery of instance
Advisory estimates extra physical writes if FSMT changed
Need to calibrate FAST_START_MTTR_TARGET
Must set STATISTICS_LEVEL = TYPICAL or ALL
Notes:
Oracle9i R2 introduces the MTTR advisory. If you are using the FAST_START_MTTR_TARGET parameter (and have the STATISTICS_LEVEL set at least at TYPICAL), Oracle will populate the v$mttr_target_advice view. Remember from chapter 1 that FAST_START_MTTR_TARGET obsoletes FAST_START_IO_TARGET and LOG_CHECKPOINT_INTERVAL. We use it to limit the number of seconds it will take to recover (all committed transactions) from a instance crash. Oracle will perform a checkpoint operation frequently enough to keep the recovery time within the number of seconds specified.
An interesting note here is that FAST_START_MTTR_TARGET should be calibrated by forcing “several instance recoveries to ensure that the time to read a redo block and the time to read and write a data block are recorded accurately.” Internal default values are used if an instance recovery has never occurred. Source: Oracle9i Database Performance Tuning Guide and Reference Release(9.2).