Interaction of Sensitivity, StatisticsWindow, and Threshold parameters

Resource-based settings use two parameters:

• Sensitivity
• StatisticsWindow

These two parameters, in conjunction with one or more thresholds, determine when an abnormal condition exists. This design provides two benefits:

• Prevents wide variations between high and low values from skewing the results, as might be the case if values were averaged.
• Provides more control over the sensitivity of the analysis. When the value of the sensitivity parameter is 0percent, one value over the threshold triggers the event. If the sensitivity parameter is 100percent, every value must be over the threshold.

The following example illustrates how these parameters work:

A SoftwareElementCheck samples a parameter such as ResponseTime every 30 seconds and compares it to the threshold, MaxResponseTime. The SoftwareElementCheck records whether the value is over the threshold. At regular intervals, determined by the correlation engine, the SoftwareElementCheck compares the number of samples that violated the threshold during the most recent window interval (StatisticsWindow) to the sensitivity parameter. If the percentage of samples in violation is higher than the sensitivity parameters, a symptom is triggered.