Normally, at this point, you have collected all the data corresponding to the hard work you did during the benchmark. It is quite important to aggregate this in a report. The report will mainly explain the investigation (why you changed some settings and so on) and expose the results of the runs.
You can, of course, ignore the useless runs (no significant change in the performance), but it is always interesting to integrate the ones corresponding to a performance boost or degradation.
The last part of the report should explain how to properly configure the server for production. It can be done inline in the report or point to another document such as a reference guide if it is about a product or a white book for an application. ...
