As you probably know, you can use Disk Watcher to collect disk performance data and use it to diagnose disk related performance problems. Typically, Disk Watcher is used for analysis focused on a particular disk unit or auxiliary storage pool (ASP) that’s having performance problems. However, this is not all Disk Watcher can do. It also has the capability to obtain data about all I/O operations being performed on the system, along with which threads or tasks are issuing the I/O operations and which objects or files are being accessed by those I/O operations. This information can be used to help understand which jobs are accessing which objects and how many I/O operations those jobs are performing on each object.
Shauna Rollings has written an article on Disk Watcher that asks and answers three questions to give you some examples:
- Which jobs or tasks issued I/O operations?
- On which objects were I/O operations performed?
- Which jobs or tasks issued I/O operations on which objects?
This kind of performance analysis could be helpful, for example, during a code update or software upgrade. Running Disk Watcher and using it to analyze a particular set of jobs both before and after a code or software update lets you see how the new code or software affects the amount of physical I/O operations occurring to particular objects and by particular jobs on your system.
This blog post is obsolete as the referenced article no longer exists.
This blog post was originally published on IBMSystemsMag.com and is reproduced here by permission of IBM Systems Media.