Expansion and Analysis of Collected Data
The collected performance data, stored in a raw and compressed binary format, must be expanded and analyzed before saving the results in the data base files used as input for all the reports and charts.
Although the analysis job indeed contains the most complex programs of GiAPA, it may the easiest part for the user, who only needs to select how much of the collected data that should be included in the analysis - the range could be anything from a few minutes to a month.
Most installations simply automate the process by scheduling the analysis to run every night, thereby processing the data from the previous day. Should it be necessary to look at data from just a few minutes ago, an additional analysis of the wanted interval can be requested interactively and be available within a few minutes.
Detailed data for jobs suffering from or potentially causing performance problems are kept for further analysis - data for other jobs are summarized by type and interval, thus reducing the space needed to store the results. For special analyses the user can request that all details should be kept.
Each set of analysis results is given a name to be used when selecting input for reporting. The name will be member name in the different files holding either details of (selected) jobs, or summaries on different levels - by collection interval, job id or name, user, or hour.
After the analysis run, which typically only takes very few minutes, all the results can be seen by requesting the many GiAPA reports and graphs available.