At this point, the only way to decrease CPU-related response time is to either use faster CPUs or reduce the SQL statement logical IO consumption (tune or balance). While additional CPUs may provide more CPU capacity, Oracle and the operating system are not able to fully take advantage of the existing four cores (for details, see the scalability discussion near the end of this chapter).

The application situation has indeed changed, as shown in Figure 9-29. First, we can see that no significant physical IO is being consumed! Thus means increasing the buffer cache had its intended affect. We were hoping for a 50% decrease in elapsed time, to around 0.316 ms/exec. What actually occurred was an elapsed time drop from 0.632 to 0.266, which is a 58% decrease in response time! So, we met and exceeded our objective. It appears the users are also able to get more work done because the SQL statement execution rate increased from 25.6 exec/sec (see Figure 9-24) to 27.1 exec/sec (Figure 9-30).

Figure 9-29. Shown is the essential application SQL information. Notice there is no physical IO consumed. Compared to Figure 9-23, the top SQL statement's elapsed time per execution improved from 0.632 ms/exec to 0.266 ms/exec, while at the same time, the number of executions during the sample interval increased from 473 to 536.