Optimizing the performance of an Oracle database, or of a system running atop an Oracle database, has long been somewhat of a black art. Database administrators have been taught to rely on utilization statistics such as buffer cache hit ratios. Developers pass on various tips and tricks for identifying and tuning poorly performing SQL. Tuning successes often come randomly, depending more on intuition, hunches, and trial-and-error than on any sort of reliable and repeatable method. Clearly, performance optimization seems like a hard problem, but is it really? In Optimizing Oracle Performance, renowned Oracle scientists Cary Millsap and Jeff Holt break new ground by describing a reliable, repeatable, and deterministic method for isolating system performance problems. Gone are the days when you had to rely on a bag of random tips and techniques. Intuition, hunches, and trial-and-error are replaced by a deterministic method that leads reliably to the root cause of any performance problem. Instance-wide statistics such as buffer cache hit ratios are thrown by the wayside, and focus is given to the one statistic that truly matters: response time as seen by the users of a system. In this book, Cary and Jeff describe their method in detail. They show you how to apply it, and they provide rigorous descriptions of why the method works. And it does work. Not only can you use the method for identifying performance bottlenecks, you can also use it to reliably predict and quantify the performance improvement that you will get from upgrade activities such as adding more CPUs, or faster CPUs, or more memory to a system.