redo log buffer
Fig. 5: Cyclic writing of the log files. © Marek Adar
Fig. 5: Cyclic writing of the log files. © Marek Adar
Due to the fact that changes are only made in the database buffer cache, these changes were lost if an instance crashed. To prevent this, changes are also logged in the redo log buffer. For each data record change, the block address, the location of the change within the block, the new value, a timestamp and a system change number (SCN) are written to this memory area. Under Oracle, all changes that belong to the same transaction are given a system change number in order to be able to identify all changes in a transaction.
Despite all this information afghanistan telegram screening being kept in the redo log buffer, any changes made would be lost in the event of an instance crash, as this memory area would itself be deleted as a consequence. Thus, there must be a mechanism to prevent data loss in the scenario just described. The only way to solve this problem is to save the contents of the redo log buffer to a permanent disk.
The contents of the redo log buffer are regularly emptied by a background process, the so-called redo log writer, and written to the redo log files. The writing process takes place every 3 seconds when the redo log buffer is one third full, before the database writer writes or when a transaction is completed with a COMMIT within the database.