To help understand the complexity of working on each step, additional columns can be helpful. For example, the table could also include the number of top SQL statements and some complexity metric. The point is that the table should encourage fruitful discussion and analysis, so an informed decision can be made about where to target the performance effort.

There are two significant differences in capturing response time information when focused on a batch process: client process time and background process time.

In vivid contrast with online activity, during batch processing, there is no think time and there can be significant client-processing time. As a result, for the total Oracle response time to equal batch process elapsed time, our response-time analysis must include client-processing time and also communication time between the client and server process. As presented in Chapter 5, this time component is captured by the SQL*Net message from client and SQL*Net more data from client wait events. If database links are involved, then don't forget to also include SQL*Net message from dblink and SQL*Net more data from dblink. When this normally useless and idle classified time is included, the batch process elapsed time will equal the total Oracle response time.