Each session should include an SM2 run, the results of which are entered in a file. With long-term measurements, the reliability of the measured data (which is dependent on the number of samples taken) can be ensured if one sample is taken per second:
sampling period (SAMPLING-PERIOD): 1000 milliseconds
monitoring cycle (OFFLINE-PERIOD): 5 minutes
analysis subinterval: 1 hour
Monitoring period: from System Ready to Shutdown
It is not necessary to evaluate each and every session. However, at least random samples in the form of daily evaluations must be made quite frequently.
Analyses over longer periods of time are the best means of evaluating trends and providing a sound basis for backing up investment decisions. The number of annual reports should be kept to a minimum.
When unusual circumstances arise, e.g. in the event of complaints, queries or changes in the configuration, it is not sufficient to analyze solely the one day in question; analysis of a second day is additionally required for purposes of comparison. If the graphs for the two analyses are placed next to one another and the figures therein are compared, relevant statements can be made much faster by noting the differences. It should be borne in mind that corresponding graphs in the two analyses might be printed using different scales, which could make the before-and-after comparison somewhat more complicated.
SM2 is also used in the support of system measurements. To this end it supplies several reports which go beyond the interests of routine system monitoring based on standard measurement criteria.