Optimizing Oracle Performance

Chapter 3. Targeting the Right Diagnostic Data

Once you have correctly targeted the user actions for which the business most needs performance improvement, it is data collection time. Diagnostic data collection is the project phase in which the typical performance analyst really begins to feel a sense of progress. There are very few arguments in Oracle literature today about how one should go about collecting performance diagnostic data. However, there should be. The way that you collect your diagnostic data has a tremendous influence over a project’s potential for success. In fact, unless you are exceptionally lucky, a performance improvement project cannot proceed beyond a botched data collection.

I hope that this chapter will surprise you. It describes a couple of very important flaws in the standard data collection procedures that are deeply institutionalized in the Oracle culture. In the hundreds of flawed Oracle performance improvement projects that my colleagues and I have helped repair, a contributing factor to failure in nearly every project was one or more errors in data collection. Unfortunately, virtually every document written about Oracle performance prior to the year 2000 leads its reader to make these errors. I believe that the commonsense examples described in this chapter will forever change your attitude toward diagnostic data collection.

Expectations About Data Collection

The whole point of data collection in a performance improvement project using Method R is ...