Tuning Recovery - Expert Oracle RAC Performance Diagnostics and Tuning - page 332

4. Time to time errors could be generated due to various reasons. A common reason is when

the FRA is not purged and is reaching its maximum limits. Space utilization could be
checked using the following scripts:

SELECT object_type,
message_type,
message_level,
reason,
suggested_action
FROM dba_outstanding_alerts;

■ if the Fra is not large enough to hold the flashback logs, then the database will overwrite flashback logs from
the earliest SCns. Consequently, the flashback database window can be shorter than the value specified by the flashback
retention target.

Note

Configuring FRA is a good practice for the overall recovery operations using RMAN. It helps to have one holistic
location for your backup and recovery process and helps in easy restore and instance/crash/media recovery operation.

Conclusion
As discussed in the early chapters of the book, the primary reason for moving to a RAC environment is availability and
then scalability. For business continuum after an instance or server or database crashes, it's important that the data
is available for users. This could be achieved by optimizing data recovery. In the case of crash or instance recovery,
after the first pass (cache recovery) is completed, users can start accessing the data. The second phase of recovery
(transaction recovery) can or could take longer and completes in the background. Depending on the business and
SLA requirements, using some of the parameters discussed could help reduce the time required to complete recovery;
however, the optimization should be measured against overall performance of the environment.
Although backup is important, equally important is the restore operation and the recovery of the database
from failures. However, not much attention is given to this area to optimize the restore operation. There are very few
parameters or areas of the system that need to be superficially tuned just to improve efficiency of the backup and
restore operations. Most of the tuning completed from other areas of the hardware infrastructure and the database
should also help improve backup and restore performance, for example, tuning the I/O subsystem, tuning the
network, tuning the database, and so forth.