As I illustrate in Chapter 1, response time optimization is a routine part of humans’ everyday lives. The foundation of response time optimization is the commonsense notion, formalized by Gene Amdahl, that improving the largest component of response time creates the greatest opportunity for response time improvement.
Recall the resource profile format, also introduced in Chapter 1, which is shown again in Example 10-1. Response time optimization is so “built into us” that most people—including users and business managers with no performance analysis training—can understand a resource profile with very little effort. Technical and non-technical audiences alike never fail to respond correctly to Example 10-1 within ten seconds of seeing the data:
I don’t know what those two “SQL*Net” things are, but whatever they are, they’re consuming nearly three quarters of the report’s total duration. What causes
message from clientand
SQL*Net more data from client?
This is exactly the right way to attack the problem. Business managers and users to whom I show this example are routinely confused about how professional performance analysts could have gone for three months believing that latch contention and CPU capacity were the root causes of this performance problem. (Example 10-1 is the same Oracle Payroll performance problem that I describe in Chapter 1.) Of course the answer is that the performance analysts on this project spent three months looking ...