From a queuing theory perspective, what really happens when service time drops is that a new response-time curve takes effect. Because the service time decreases, with no load on the system and therefore no queuing, the response time with minimal arrivals is less. So, the curve has shifted down. But it gets better. Because each transaction server (for example, a CPU core) can process each arrival quicker, it can process more arrivals per unit of time before queuing sets in, which shifts the graph to the right. So, tuning shifts the response-time curve down and also to the right!

Figure 9-17 graphically shows how tuning can affect a system. Starting at point A, the performance is unacceptable and highly variable. By tuning the application, Oracle, or the operating system, the response time decreases (that is, improves), and the system is operating at point B. However, now the administrators have a choice. By controlling the workload (the arrival rate), they can allow more work to flow the system without affecting the response time. Point C shows this negligible affect on response time by allowing the arrival rate to increase. So again, tuning provides the performance analyst with several options: decreased response time, increased workload, or a managed combination of both!

Figure 9-17. Shown is the response time effect of tuning. By tuning, a new response-time curve takes effect (dotted line), and the response time drops from point A to point B. By controlling the workload, performance can remain at point B or by allowing the workload to increase to point C, the system can still maintain both improved response time and an increased workload.