Figure 5-12. This OraPub 120-second interval response time report clearly shows Oracle processes are suffering from intense cache buffer chain latch contention. A response time-focused analysis will work on reducing the response time, not only the wait time or the service time.

Most Oracle DBAs will look at the 168 seconds of cache buffer chains time and focus on reducing the wait time, without regard to the effect on service time or the Oracle workload (not shown). Suppose they decide to increase the spin count. While this may indeed reduce latch sleep time, it will very likely also increase CPU consumption! If the DBA is lucky, this may result in a net response time decrease. While we appreciate luck and take it when it comes, that's no way to optimize systems or build a career. The solution here is to focus on reducing response time, because that is closer to what the user experiences.

As Oracle systems become increasingly complex, the time attributed to the Oracle database server is becoming less significant. Said another way, Oracle's time contribution toward what a user experiences is decreasing. While that may sound fine, this also means Oracle's time contribution is becoming more and more insignificant. This makes our job more difficult and our impact less significant, because our optimization efforts will have less and less effect on the total user experience. (This is also not great for job security!)