Databases Reference
In-Depth Information
In step 9, we query the
V$INSTANCE_RECOVERY
view to obtain an estimate of recovery time,
collecting data required to choose the correct values for the parameters shown earlier. In the
following screenshot, you can observe the output of this query:
You can find the description for each field in
V$INSTANCE_RECOVERY
view in
Appendix A
,
Dynamic Performance Views
.
There's more...
In this recipe, we have seen how different parameters can influence the checkpoint behavior.
The checkpoint queue is a list of dirty buffers (that is, blocks in the buffer cache modified and
not already written to the disk) waiting to be written to the disk by the
DBWR
processes.
There is a trade-off between a short checkpoint queue that ensures faster recovery times in
case of instance crash, and a long checkpoint queue which avoids frequent
DBWR
writes that
can affect performance.
As always, we need to evaluate the ideal configuration to satisfy our Service Level
Agreements. If we assure the maximum recovery time, we will shorten the checkpoint queue,
resulting in some more
DBWR
writes.
