The profiling service generates information related to O3 Server as well as to each cube published in it. This information is saved in different log files, whose structures and characteristics are presented below.
The generation of log information by the profiling service can be performed in two different ways: sampling certain indicators at regular time frames, or in the face of any given event, such as a query being made, a user logging in or logging out, or a failed connection to the server.
In the first case, as the time frame defined in the setup goes by, a record is generated in some of the log files (O3CubeService, Sessions y Cubes), with the current values of the defined indicators (for instance: cache memory used at that time, number of open sessions at the time, etc.).
For this information to be used in the building of a cube, and to eventually enable the comparison among different days, weeks, months, etc., the samples must always be taken at the same time frame in the day (that is to say that the beginning and ending of the sampling must occur at the same time) and with the same frequency. If any of these parameters are modified, it is advisable to delete previous data from the log files.
In the second case, each time any user makes, let us say, a query on O3 Server, a record is generated in some of the files (Audit and Queries) with information related to the performed query (query time in milliseconds, percentage of correct answers in the cache for that query, etc.) or with the values of some indicators at that time. Something similar happens in the case of a log in event.