Figure 4-22 is a typical iostat report for a very small IO subsystem. It is common for a thousand lines to be returned. When studying the IO subsystem, it can be a good idea to load the iostat data into Oracle and filter on the response time (await, svctm) and utilization (%util) columns to find offensive IO patterns.

The iostat report can be initially overwhelming, but you need to look at only a few columns to determine if there is a performance problem. The manual pages are also very comprehensive, so you can refer to them. Here are brief descriptions of the key columns:

* Device: This is the device name, which can be just about anything these days. It could a volume, a volume group, a RAID array, an actual physical disk, a disk partition, or something else. The key is to first determine if Oracle is experiencing any IO response time issues. Then, in the iostat report, look for active yet poorly performing devices. If you find any, present both your Oracle and iostat analyses to your IO administrator.