||Oracle Tips by Burleson
Oracle and Bandwidth Saturation
A sub-optimal disk configuration can make even a well-tuned Oracle
RAC database slow to a crawl. An Oracle clustered environment has
multiple database instances all sharing the same set of data files.
MetaLink Note: 62172.1 notes that the top remedy for increasing DBWR
throughput is increasing the speed of the device.
This example from an AWR report from $ORACLE_HOME/rdbms/admin/awrrpt.sql
shows an I/O bound database:
Top 5 Timed Events
~~~~~~~~~~~~~~~~~~ % Total
Event Waits Time (s) Ela Time
-------------------------- --------- ----------- --------
db file sequential read 2,598 7,146 48.54
db file scattered read 25,519 3,246 22.04
library cache load lock 673 1,363 9.26
CPU time 1,154 7.83
log file parallel write 19,157 837 5.68
Since most DBAs are too busy to run disk performance scripts on
every node of a large RAC cluster, they can look to the web for
examples of scripts. Another source is Mike Ault’s book Oracle Disk
I/O Tuning (2004, Rampant TechPress), which also provides a plethora
of scripts and examples.
For the vast majority of non-scientific systems, the primary
bottleneck is disk I/O. Back in the days before RAID and giant
db_cache_size, the DBA had to manually load balance the disk I/O
sub-system to relieve contention on the disks and the disk
controllers. The next section explores this topic.
The above book excerpt is from:
& Grid Tuning with Solid State Disk
Expert Secrets for High Performance Clustered Grid Computing
Mike Ault, Donald K. Burleson