If we take CPU load as an example metric, Analyzer uses the following formula to determine load:
1.0 (CPU Idle Time / CPU Total Time).
CPU Total Time is determined by the polling cycle setting. So CPU Load on Server A over the past five minutes may be represented differently on a performance graph, depending on the polling during. However, the average load over those five minutes is the same. As Analyzer begins to aggregate data (keeping the average and peaks), differences in polling cycle are averaged out and data is represented the same regardless of the polling cycle setting. Analyzer begins aggregating performance data after 72 hours (aggregating to 1 hour granularity) and then again after two weeks (to one day granularity).
As mentioned, aggregated data contains average and peak values. On the performance graph, one line represents the average performance value and the other represents the highest reported value for the time period. As an example, a CPU performance graph showing a given system¹s performance from 10 days ago will show one line which represents that day's average performance value and the other line represents that day's highest recorded value.
