This example demonstrates that a relatively simple ORTA, combined with a quick 3-circle analysis, results in a spot-on diagnosis, a number of solution possibilities, and graphics to help everyone understand the situation.

Remember your school report card? It may have included your current term's grade point average and also an accumulated grade point average. Both statistics are important, yet based on a different data set. Oracle reports based on performance views can also yield current or single-term data (think interval) or accumulated data (think since instance start). For example, running a quick SQL statement referencing an Oracle performance view, such as v$sysstat, does not necessarily represent what is currently occurring, what just occurred, or what has recently occurred. It represents what has been occurring since the Oracle instance started! Therefore, unless you are careful, you can be deceived.

Suppose an Oracle instance has been running for two months and you wanted to know the cache hit ratio for the buffer cache. Querying v$sysstat once would give you the average cache hit ratio since the instance has started. If you wanted to know the cache hit ratio over a specific period, you would need to have beginning and ending values. Think in terms of Statspack and AWR. Before you run the report, you must set both the start and end points. This provides an interval, or period of time, so your calculations are relevant to the time period desired. I call the interval calculations the blue line, and the simple v$sysstat query the red line, as illustrated in Figure 1-5.