Redo Logs contain Redo Entries; basically this is data which can be used to replay changes made to the database. Both the Oracle background processes and user sessions are constantly writing to the redo logs via lgwr as transactions occur. Writing occurs via the Redo Buffer, and if insufficient space is available in the buffer, process waiting times will increment.

The metric value crucial to monitoring the effectiveness of the Redo Buffer is "Redo log space requests" which may be found in v$sysstat.

This value should be zero, indicating that processes have sufficient redo buffer space. Once this starts to increase, remedial action must be taken before performance degradation becomes significant. As ever, this involves an init.ora parameter.

The log_buffer parameter in init.ora should be increased. Remember that the increased value should be a multiple of the database block size. Also remember to check on the state of your free memory, otherwise you'll be trading process waiting time for paging and/or swapping.