Most Oracle installations do not maintain the default redo log file
organization because it becomes a single point of failure. If one of your redo log files becomes corrupted, the system stops working. To address this problem, Oracle provides you with a way to multiplex redo log files.
Multiplexed redo log files
The following diagram explains multiplex redo log files.
This is member 1 of group 1 stored on disk 1.
This is member 2 of group 1 stored on disk 2.
This is member 1 of group 2 stored on disk 1.
This is member 2 of group 2 stored on disk 2.
LGWR will write to members 1 & 2 of group 1. When a switch occurs, LGWR will write to members 1 & 2 of group 2.
If disk 1 fails, LGWR will continue writing to disk 2 without database failure.
If disk 2 fails, <tt>LGWR</tt> will continue writing to disk 1 without database failure.
There must be at least 2 members of the same size in each group. Each group should contain the same number of members.
There must be at least 2 members of the same size in each group. Each group should contain the same number of members.
Let us take the above example and add a new redo log group. There are two init.ora parameters that you should be aware of. First, you need to make sure MAXLOGFILES is set so that you can add a new log file group. If we have a third disk drive and want to add a third member to each group, we must check MAXLOGMEMBERS. So lets add a third redo log group. We will use Server Manager to do this.
svrmgr> alter database add logfile group 3
('c:\oracle8\database\g3_loga.ora',
'd:\oracle8\database\g3_logb.ora')
The results of this are shown by the following MouseOver:
This is member 1 of group 1 stored on disk 1.
This is member 1 of group 1 stored on disk 1.
This is member 1 of group 2 stored on disk 1.
This is member 2 of group 2 stored on disk 2.
This is member 2 of group 2 stored on disk 2.
This is member 2 of group 3 stored on disk 2.
This is member 2 of group 3 stored on disk 2.
If disk 1 fails, <tt>LGWR</tt> will continue writing to disk 2 without database failure.
If disk 1 fails, <tt>LGWR</tt> will continue writing to disk 2 without database failure.
There must be at least 2 members of the same size in each group. Each group should contain the same number of members.
There must be at least 2 members of the same size in each group. Each group should contain the same number of members
There must be at least 2 members of the same size in each group. Each group should contain the same number of members.