Optimizing Oracle Performance by Cary V. Millsap
This text dispenses with database-centric metrics and instead shows how to use Oracle's response time statistics to diagnose performance problems in an Oracle application system. Former VP of Oracle's System Performance Group, Cary Millsap, clearly shows how eliminating requests for unneeded work improves performance dramatically. He also shows how the mathematical discipline of queueing theory can be combined with response time metrics to accurately predict the impact of system changes and upgrades. This text clearly and concisely explains how to use Oracle's response time statistics to diagnose and repair performance problems. It also shows how "queueing theory" can be applied to response time statistics to predict the impact of upgrades and other system changes. The book aims to eliminate the time-consuming, trial-and-error guesswork inherent in most conventional approaches to tuning. You can determine exactly where a system's performance problem is, and with equal importance,where it is not, in just a few minutes - even if the problem is several years old. It also intends to cut a path through the complexity of current tuning methods, and streamline an approach that focuses on optimization techniques that any DBA can use quickly and successfully to make noticeable - even dramatic - improvements. For example, the one thing database users care most about is response time. Naturally, DBAs focus much of their time and effort towards improving response time. But it is entirely too easy to spend hundreds of hours to improve important system metrics such as hit ratios, average latencies, and wait times, only to find users are unable to perceive the difference. And an expensive hardware upgrade may not help either.