Document Type | Technical Information
Category | Backup/Recovery
Document Number | TBATI028
Overview
If the amount of rollback to be performed is confirmed to be large through V$TRANSACTION and waiting is difficult, requiring an abnormal shutdown of the DB, this guide explains how to reduce the cache recovery time that occurs during the next boot.
Method
Before shutting down the DB, perform a checkpoint to write the contents of the changed buffer cache to disk, thereby reducing the amount of recovery needed.
SQL> ALTER SYSTEM CHECKPOINT;
A checkpoint flushes dirty blocks existing in the DB buffer cache to disk based on the point in time when it is performed.
It is assumed that data blocks before that point do not need to be recovered, so the amount of recovery is reduced. (However, this condition requires that no load is applied during the checkpoint process.)
Note
Dirty block: A block in the buffer cache whose contents have been changed but have not yet been synchronized with the disk